Wall of love for The Gen Academy

Submit your testimonialBuild your own wall? It's free šŸ‘‰
Rajmohan Dhanapal
Great learning session on GenAI Governance, Security, and Observability! I had the opportunity to attend an insightful presentation that explored some of the real-world challenges organizations need to think about as they move GenAI applications from experimentation into production. One topic that particularly stood out to me was how LLM/GenAI observability and evaluation can be aligned with the OWASP Top 10 risks for GenAI, including: šŸ”¹ Prompt Injection šŸ”¹ Sensitive Information Disclosure šŸ”¹ Data/Model Poisoning šŸ”¹ Improper Output Handling šŸ”¹ Excessive Agency šŸ”¹ System Prompt Leakage šŸ”¹ Vector/Embedding Weakness šŸ”¹ Misinformation & Hallucinations As we build more RAG systems and AI Agents, it is becoming increasingly clear that getting a good response from an LLM is only part of the solution. We also need to think about evaluation, security, governance, monitoring, and trust throughout the AI application lifecycle. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for putting together this thoughtful session. Also appreciate Nick Nolan, for the insightful discussion and knowledge sharing. Great session and lots of valuable takeaways! Aadhil A. Majeed John Ninan Aswath Harish Jayaprakash Gautham Guganesh Avantika Dharmagadda, PMP #GenerativeAI #GenAI #AIAgents #RAG #LLM #AIGovernance #AISecurity #LLMOps #OWASP #ResponsibleAI
Great learning session on GenAI Governance, Security, and Observability! I had the opportunity to attend an insightful presentation that explored some of the real-world challenges organizations need to think about as they move GenAI applications from experimentation into production. One topic that particularly stood out to me was how LLM/GenAI observability and evaluation can be aligned with the OWASP Top 10 risks for GenAI, including: šŸ”¹ Prompt Injection šŸ”¹ Sensitive Information Disclosure šŸ”¹ Data/Model Poisoning šŸ”¹ Improper Output Handling šŸ”¹ Excessive Agency šŸ”¹ System Prompt Leakage šŸ”¹ Vector/Embedding Weakness šŸ”¹ Misinformation & Hallucinations As we build more RAG systems and AI Agents, it is becoming increasingly clear that getting a good response from an LLM is only part of the solution. We also need to think about evaluation, security, governance, monitoring, and trust throughout the AI application lifecycle. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for putting together this thoughtful session. Also appreciate Nick Nolan, for the insightful discussion and knowledge sharing. Great session and lots of valuable takeaways! Aadhil A. Majeed John Ninan Aswath Harish Jayaprakash Gautham Guganesh Avantika Dharmagadda, PMP #GenerativeAI #GenAI #AIAgents #RAG #LLM #AIGovernance #AISecurity #LLMOps #OWASP #ResponsibleAI

Arjun D
This week, I built AI workflows in two completely different ways ā€” with code and with no-code. šŸ”§ Same goal: build AI that actually does useful work. Two very different approaches. CODE ā†’ FinRAG I built a RAG system in Python using LangChain that answers questions over real SEC filings. ā†’ Custom chunking ā†’ Reranking ā†’ Retrieval strategy ā†’ Full evaluation harness It gave me deep control over every part of the pipeline. NO-CODE ā†’ n8n I built an AI workflow visually by connecting a trigger, an AI Agent, and a Gemini model. Input: a paragraph. Output: a clean, structured summary. From idea to working automation in minutes. Building both back-to-back made one thing very clear: Agent concepts make much more sense when you see them working. In n8n, memory and tools aren't just abstract concepts ā€” they're components you connect when an agent needs to remember something or take an action. And the biggest lesson? Code gives you control. No-code gives you speed. FinRAG needed control because retrieval, chunking, reranking, and evaluation required deeper customization. The n8n workflow needed speed ā€” and no-code was perfect for that. They aren't competitors. They're different tools for different jobs. The real skill is knowing which one to reach for ā€” and when. Grateful to Aishwarya Srinivasan and Arvind Narayanamurthy for continuing to push us to build across the entire AI stack, not just the parts we're already comfortable with. #AgenticAI #n8n #RAG #LangChain #AI #NoCode #BuildInPublic #TheGenAcademy
This week, I built AI workflows in two completely different ways ā€” with code and with no-code. šŸ”§ Same goal: build AI that actually does useful work. Two very different approaches. CODE ā†’ FinRAG I built a RAG system in Python using LangChain that answers questions over real SEC filings. ā†’ Custom chunking ā†’ Reranking ā†’ Retrieval strategy ā†’ Full evaluation harness It gave me deep control over every part of the pipeline. NO-CODE ā†’ n8n I built an AI workflow visually by connecting a trigger, an AI Agent, and a Gemini model. Input: a paragraph. Output: a clean, structured summary. From idea to working automation in minutes. Building both back-to-back made one thing very clear: Agent concepts make much more sense when you see them working. In n8n, memory and tools aren't just abstract concepts ā€” they're components you connect when an agent needs to remember something or take an action. And the biggest lesson? Code gives you control. No-code gives you speed. FinRAG needed control because retrieval, chunking, reranking, and evaluation required deeper customization. The n8n workflow needed speed ā€” and no-code was perfect for that. They aren't competitors. They're different tools for different jobs. The real skill is knowing which one to reach for ā€” and when. Grateful to Aishwarya Srinivasan and Arvind Narayanamurthy for continuing to push us to build across the entire AI stack, not just the parts we're already comfortable with. #AgenticAI #n8n #RAG #LangChain #AI #NoCode #BuildInPublic #TheGenAcademy

Selva Mohandoss
InterviewDNA ā€” a personalized interview coach that reads your resume and a target job description, finds the real gaps, and runs an adaptive mock interview that changes in real time based on how you actually answer. Not a chatbot. Not a static question bank. An actual agentic system. **What that meant in practice:** ā†’ A LangGraph state machine ā€” not the LLM ā€” decides what happens after every answer: probe deeper, raise the difficulty, coach a real knowledge gap, or move on. The model reasons and generates; the code decides and controls. ā†’ Hybrid search for grounding ā€” vector search (Pinecone) + keyword search (BM25) fused together, then reranked with a cross-encoder. Because semantic search alone misses things a plain keyword match would catch, and vice versa. ā†’ A genuine tool-use agent ā€” when it detects a knowledge gap, it decides whether to search its own indexed material or the live web, then grounds its coaching in whatever it actually finds. No hallucinated explanations. ā†’ An evals harness, not just unit tests ā€” because "does my code run" and "is the model's output actually good" are completely different questions. I built LLM-as-judge checks specifically to catch things keyword matching can't: hallucination, unfaithful grounding, subtle scoring bugs. I chased a production bug for hours that turned out to be blocking async code freezing an entire FastAPI server. I hit a hard memory ceiling on free-tier hosting and had to actually understand what PyTorch's CUDA-bundled wheel was doing to my container's RAM. I found a race condition between writing to a vector database and immediately querying it. I debugged a model provider silently deprecating the exact model I'd built against. Every one of those was a systems engineering problem wearing an AI costume. That's the actual skill gap between "using AI" and "engineering AI systems" ā€” and it's a lot closer to backend engineering than the hype suggests. Fully self-hosted, fully free to run end to end: local LLM, hybrid retrieval, structured evals, real deployment debugging included. Happy to talk through any piece of this ā€” the agentic loop, the retrieval design, or the very unglamorous production bugs that taught me the most. github: lnkd.in/gvJuZq8n Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy pushing hands-on building over passive learning. #AIEngineering #LangGraph #RAG #BuildInPublic
InterviewDNA ā€” a personalized interview coach that reads your resume and a target job description, finds the real gaps, and runs an adaptive mock interview that changes in real time based on how you actually answer. Not a chatbot. Not a static question bank. An actual agentic system. **What that meant in practice:** ā†’ A LangGraph state machine ā€” not the LLM ā€” decides what happens after every answer: probe deeper, raise the difficulty, coach a real knowledge gap, or move on. The model reasons and generates; the code decides and controls. ā†’ Hybrid search for grounding ā€” vector search (Pinecone) + keyword search (BM25) fused together, then reranked with a cross-encoder. Because semantic search alone misses things a plain keyword match would catch, and vice versa. ā†’ A genuine tool-use agent ā€” when it detects a knowledge gap, it decides whether to search its own indexed material or the live web, then grounds its coaching in whatever it actually finds. No hallucinated explanations. ā†’ An evals harness, not just unit tests ā€” because "does my code run" and "is the model's output actually good" are completely different questions. I built LLM-as-judge checks specifically to catch things keyword matching can't: hallucination, unfaithful grounding, subtle scoring bugs. I chased a production bug for hours that turned out to be blocking async code freezing an entire FastAPI server. I hit a hard memory ceiling on free-tier hosting and had to actually understand what PyTorch's CUDA-bundled wheel was doing to my container's RAM. I found a race condition between writing to a vector database and immediately querying it. I debugged a model provider silently deprecating the exact model I'd built against. Every one of those was a systems engineering problem wearing an AI costume. That's the actual skill gap between "using AI" and "engineering AI systems" ā€” and it's a lot closer to backend engineering than the hype suggests. Fully self-hosted, fully free to run end to end: local LLM, hybrid retrieval, structured evals, real deployment debugging included. Happy to talk through any piece of this ā€” the agentic loop, the retrieval design, or the very unglamorous production bugs that taught me the most. github: lnkd.in/gvJuZq8n Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy pushing hands-on building over passive learning. #AIEngineering #LangGraph #RAG #BuildInPublic

Srinivas Simhan
Week 1 of Mastering Agentic AI started this weekend, a six-week cohort on Maven taught by Aishwarya Srinivasan and Arvind Narayanamurthy through The Gen Academy. The first session had over 350 people on the call. The chat filled with locations as people joined, North America, India, Europe, Australia, Singapore, South Africa, Dubai. One person was grabbing coffee at 1am local time to attend live! The sheer range of people deciding agentic AI is worth their Saturday says something before any curriculum does. I've followed Aishwarya's work for a while, and her approach to teaching AI, grounded in building rather than theory, is a big part of why I signed up. The program runs live sessions, guest lectures across time zones, and group builds. I'll be sharing what I learn and what I build as the six weeks go on, as the reading list alone has already changed what my browser tabs look like. #AgenticAI #AI #BuildingInPublic
Week 1 of Mastering Agentic AI started this weekend, a six-week cohort on Maven taught by Aishwarya Srinivasan and Arvind Narayanamurthy through The Gen Academy. The first session had over 350 people on the call. The chat filled with locations as people joined, North America, India, Europe, Australia, Singapore, South Africa, Dubai. One person was grabbing coffee at 1am local time to attend live! The sheer range of people deciding agentic AI is worth their Saturday says something before any curriculum does. I've followed Aishwarya's work for a while, and her approach to teaching AI, grounded in building rather than theory, is a big part of why I signed up. The program runs live sessions, guest lectures across time zones, and group builds. I'll be sharing what I learn and what I build as the six weeks go on, as the reading list alone has already changed what my browser tabs look like. #AgenticAI #AI #BuildingInPublic

Rajmohan Dhanapal
Starting a New Learning Journey: Mastering Agentic AI: Certification by The Gen Academy Excited to start the Mastering AI Agents 6 Weeks cohort with Aishwarya Srinivasan , Arvind Narayanamurthy and The Gen Academy !!! As I continue expanding my journey from Data Engineering into AI/ML and Agentic AI, Iā€™m looking forward to moving beyond understanding individual AI concepts toward learning how to design and build production-grade AI applications. We kicked things off with a strong foundation, covering: šŸ”¹ Evolution of LLMs šŸ”¹ Determinism vs. Non-Determinism šŸ”¹ LLM Architecture & Model Landscape šŸ”¹ Prompt Foundations & Techniques šŸ”¹ AI Agents, Harnesses and Loops šŸ”¹ The growing Agentic AI technology stack One thing is already clear: this is going to be a marathon, not a sprint. šŸƒā™‚ļø There is a huge difference between experimenting with an LLM and building an AI application that is reliable, observable, secure, and production-ready. Iā€™m especially excited to learn how all the pieces of the Agentic AI stack come together to bridge that gap. Looking forward to learning, experimenting, and most importantly, building production-grade applications alongside this cohort over the coming weeks. The journey continues! šŸš€ John Ninan Gnana Sudheer Gavarraju Aswath Harish Jayaprakash Gautham Guganesh #AgenticAI #AIAgents #GenerativeAI #LLM #ArtificialIntelligence #PromptEngineering #AIEngineering #MachineLearning #DataEngineering #ContinuousLearning
Starting a New Learning Journey: Mastering Agentic AI: Certification by The Gen Academy Excited to start the Mastering AI Agents 6 Weeks cohort with Aishwarya Srinivasan , Arvind Narayanamurthy and The Gen Academy !!! As I continue expanding my journey from Data Engineering into AI/ML and Agentic AI, Iā€™m looking forward to moving beyond understanding individual AI concepts toward learning how to design and build production-grade AI applications. We kicked things off with a strong foundation, covering: šŸ”¹ Evolution of LLMs šŸ”¹ Determinism vs. Non-Determinism šŸ”¹ LLM Architecture & Model Landscape šŸ”¹ Prompt Foundations & Techniques šŸ”¹ AI Agents, Harnesses and Loops šŸ”¹ The growing Agentic AI technology stack One thing is already clear: this is going to be a marathon, not a sprint. šŸƒā™‚ļø There is a huge difference between experimenting with an LLM and building an AI application that is reliable, observable, secure, and production-ready. Iā€™m especially excited to learn how all the pieces of the Agentic AI stack come together to bridge that gap. Looking forward to learning, experimenting, and most importantly, building production-grade applications alongside this cohort over the coming weeks. The journey continues! šŸš€ John Ninan Gnana Sudheer Gavarraju Aswath Harish Jayaprakash Gautham Guganesh #AgenticAI #AIAgents #GenerativeAI #LLM #ArtificialIntelligence #PromptEngineering #AIEngineering #MachineLearning #DataEngineering #ContinuousLearning

Raghu Veer M V
Yesterday, I embarked on the marathon learning journey of Mastering Agentic AI with The Gen Academy, led by Aishwarya Srinivasan and Arvind Narayanamurthy. Coming from a predominantly business and transformation background rather than a deep technical one, I was initially curious about how accessible this learning experience would be. What stood out immediately was Aishwarya's exceptional ability to simplify complex AI concepts into practical, easy-to-understand insights. The introduction to vibe coding was particularly inspiring, making AI development feel far more approachable and motivating than I had imagined. I'm excited about the journey ahead and look forward to expanding my understanding of Agentic AI, building hands-on skills, and exploring the possibilities this technology creates. Thank you, Aishwarya and Aravind, for creating such an engaging and inclusive learning experience. Looking forward to learning much more over the coming weeks! #AgenticAI #GenerativeAI #AITransformation #ContinuousLearning #GenAcademy #Aishwarya #Aravind
Yesterday, I embarked on the marathon learning journey of Mastering Agentic AI with The Gen Academy, led by Aishwarya Srinivasan and Arvind Narayanamurthy. Coming from a predominantly business and transformation background rather than a deep technical one, I was initially curious about how accessible this learning experience would be. What stood out immediately was Aishwarya's exceptional ability to simplify complex AI concepts into practical, easy-to-understand insights. The introduction to vibe coding was particularly inspiring, making AI development feel far more approachable and motivating than I had imagined. I'm excited about the journey ahead and look forward to expanding my understanding of Agentic AI, building hands-on skills, and exploring the possibilities this technology creates. Thank you, Aishwarya and Aravind, for creating such an engaging and inclusive learning experience. Looking forward to learning much more over the coming weeks! #AgenticAI #GenerativeAI #AITransformation #ContinuousLearning #GenAcademy #Aishwarya #Aravind

Utthra Kaushik
This post was drafted by an agent I built this week. šŸ¤– Week 3 of the Mastering Agentic AI bootcamp, and I shipped my first end-to-end agentic system: a multi-agent pipeline with a human in the loop. The problem was personal. I want to keep building in public, but writing a post every week eats 2ā€“3 hours of re-reading slides and staring at a blank page. So I made that the project. Meet the Build-in-Public Content Agent. It ingests a week's course materials, my own notes, and my project repo, then pairs what I learned with what I actually built. A planner picks the themes, a generator drafts for LinkedIn and my GitHub Pages blog, and a critic scores every draft for grounding and tone ā€” looping back until it clears the bar. Nothing publishes without me. A LangGraph interrupt() halts the graph and hands me the draft. Reads are autonomous; writes wait for a human. The lesson that stuck: the hard part of agents isn't the prompts. It's control flow, state, and drawing the line between what the agent does alone and what needs a human. (Yes ā€” it wrote this too.) Code + a 5-min demo in the comments šŸ‘‡ What weekly workflow would you turn into an agent first? Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for creating a course that just gets better every week! #AgenticAI #LangGraph #BuildInPublic #LLM
This post was drafted by an agent I built this week. šŸ¤– Week 3 of the Mastering Agentic AI bootcamp, and I shipped my first end-to-end agentic system: a multi-agent pipeline with a human in the loop. The problem was personal. I want to keep building in public, but writing a post every week eats 2ā€“3 hours of re-reading slides and staring at a blank page. So I made that the project. Meet the Build-in-Public Content Agent. It ingests a week's course materials, my own notes, and my project repo, then pairs what I learned with what I actually built. A planner picks the themes, a generator drafts for LinkedIn and my GitHub Pages blog, and a critic scores every draft for grounding and tone ā€” looping back until it clears the bar. Nothing publishes without me. A LangGraph interrupt() halts the graph and hands me the draft. Reads are autonomous; writes wait for a human. The lesson that stuck: the hard part of agents isn't the prompts. It's control flow, state, and drawing the line between what the agent does alone and what needs a human. (Yes ā€” it wrote this too.) Code + a 5-min demo in the comments šŸ‘‡ What weekly workflow would you turn into an agent first? Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for creating a course that just gets better every week! #AgenticAI #LangGraph #BuildInPublic #LLM

Ravi Kiran Uppalapati
šŸš€ Week 3 Project Done ā€” Bulls & Cows meets LangGraph! For this week's project, I built an Agent vs Human Bulls and Cows game powered by LangGraph and Streamlit ā€” and I'm genuinely proud of how it came together. šŸŽÆ What it does: An AI Opponent Agent uses LangGraph to intelligently guess your secret number, while a Coach Agent watches every move, tracks the full game history ā€” guesses, feedback, candidate counts ā€” and gives you real-time reasoning notes to help you play better. Both agents can be powered by a local Ollama model or hosted LLMs like Gemini or Nebius. Under the hood: deterministic tools handle the bulls/cows scoring and candidate filtering so the game stays correct, while the LLM makes it conversational and engaging. LangSmith tracing ties it all together for visibility into every agent turn. šŸ”§ Stack: Python Ā· LangGraph Ā· Streamlit Ā· Ollama Ā· LangSmith Ā· Nebius/Gemini šŸ‘‰ lnkd.in/eCea8efg Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy ā€” your teaching pushed me to think in graphs and agents, not just scripts. The way you broke down LangGraph concepts week by week made this project feel achievable. And to my amazing cohort ā€” watching your projects, your questions in sessions, and your encouragement kept me going when the agent wasn't cooperating šŸ˜…. You all set the bar high and inspired me to clear it. On to Week 4! šŸ’Ŗ #LangGraph #AI #AgentAI #Python #Streamlit #LLM #BuildingInPublic #Week3
šŸš€ Week 3 Project Done ā€” Bulls & Cows meets LangGraph! For this week's project, I built an Agent vs Human Bulls and Cows game powered by LangGraph and Streamlit ā€” and I'm genuinely proud of how it came together. šŸŽÆ What it does: An AI Opponent Agent uses LangGraph to intelligently guess your secret number, while a Coach Agent watches every move, tracks the full game history ā€” guesses, feedback, candidate counts ā€” and gives you real-time reasoning notes to help you play better. Both agents can be powered by a local Ollama model or hosted LLMs like Gemini or Nebius. Under the hood: deterministic tools handle the bulls/cows scoring and candidate filtering so the game stays correct, while the LLM makes it conversational and engaging. LangSmith tracing ties it all together for visibility into every agent turn. šŸ”§ Stack: Python Ā· LangGraph Ā· Streamlit Ā· Ollama Ā· LangSmith Ā· Nebius/Gemini šŸ‘‰ lnkd.in/eCea8efg Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy ā€” your teaching pushed me to think in graphs and agents, not just scripts. The way you broke down LangGraph concepts week by week made this project feel achievable. And to my amazing cohort ā€” watching your projects, your questions in sessions, and your encouragement kept me going when the agent wasn't cooperating šŸ˜…. You all set the bar high and inspired me to clear it. On to Week 4! šŸ’Ŗ #LangGraph #AI #AgentAI #Python #Streamlit #LLM #BuildingInPublic #Week3

Avishek Bhattacharjee
Built an AI-Powered Jira Sprint Reporting Agent this week. šŸš€ --------------------------------------------------------- The goal was simple: Instead of manually reviewing Jira boards every week, let an agent fetch sprint data, analyze project health, identify blockers, and generate an executive-ready status report. How it works ** Connects directly to Jira Cloud using the latest Atlassian REST APIs ** Fetches: - Active sprint - Sprint issues - Issue details - JQL search results Historical sprint information šŸ”¹ Normalizes and enriches the data šŸ”¹ Uses an LLM as the reasoning engine to: Summarize sprint progress 1. Identify blockers and delivery risks 2. Highlight sprint trends 3. Recommend actionable next steps 4. Predict areas that need attention in the next sprint The interesting part šŸ‘‡ The agent doesn't stop after generating a report. It stores weekly snapshots (metrics, blockers, risks, recommendations) and validates itself when the next sprint starts. That means it can answer questions like: āœ… Which issues have been blocked for more than one sprint? āœ… Were the predicted risks actually realized? āœ… Did blocker resolution improve sprint velocity? āœ… How accurate were last week's recommendations? One takeaway from building this: Not everything needs AI. Fetching data, calculating velocity, identifying blockers, and comparing sprints can all be implemented with deterministic business rules. The LLM becomes valuable when it's time to transform structured project data into executive summaries, explain risks, recommend actions, and answer natural language questions. That's where AI complements analytics rather than replacing it. Curious to hear how others are approaching AI-powered project intelligence beyond traditional dashboards. Aishwarya Arvind Repo: lnkd.in/gaFBMK7B #LangGraph #LangChain #Jira #Atlassian #Python #LLM #ProjectManagement #Automation
Built an AI-Powered Jira Sprint Reporting Agent this week. šŸš€ --------------------------------------------------------- The goal was simple: Instead of manually reviewing Jira boards every week, let an agent fetch sprint data, analyze project health, identify blockers, and generate an executive-ready status report. How it works ** Connects directly to Jira Cloud using the latest Atlassian REST APIs ** Fetches: - Active sprint - Sprint issues - Issue details - JQL search results Historical sprint information šŸ”¹ Normalizes and enriches the data šŸ”¹ Uses an LLM as the reasoning engine to: Summarize sprint progress 1. Identify blockers and delivery risks 2. Highlight sprint trends 3. Recommend actionable next steps 4. Predict areas that need attention in the next sprint The interesting part šŸ‘‡ The agent doesn't stop after generating a report. It stores weekly snapshots (metrics, blockers, risks, recommendations) and validates itself when the next sprint starts. That means it can answer questions like: āœ… Which issues have been blocked for more than one sprint? āœ… Were the predicted risks actually realized? āœ… Did blocker resolution improve sprint velocity? āœ… How accurate were last week's recommendations? One takeaway from building this: Not everything needs AI. Fetching data, calculating velocity, identifying blockers, and comparing sprints can all be implemented with deterministic business rules. The LLM becomes valuable when it's time to transform structured project data into executive summaries, explain risks, recommend actions, and answer natural language questions. That's where AI complements analytics rather than replacing it. Curious to hear how others are approaching AI-powered project intelligence beyond traditional dashboards. Aishwarya Arvind Repo: lnkd.in/gaFBMK7B #LangGraph #LangChain #Jira #Atlassian #Python #LLM #ProjectManagement #Automation

Sham Sundar Hassan Chikkegowda
Is your AI actually remembering anything, or is it just repeating itself? In B2B work, 'memory' isn't just about recalling factsā€”itā€™s about preventing context from poisoning the next decision. #GenAcademy had Rudraj Mehta (Product Lead at #Mem0) walk through how memory algorithms actually work ā€” the session demo was a restaurant chatbot that remembers you like pasta and avoids mushrooms. Simple example, makes the architecture click. But the three phases he described ā€” extract, consolidate, retrieve ā€” are exactly what #B2BAI teams keep rebuilding from scratch when their agents lose context between sales calls. The things #RAG and Claude alone don't cover in a cross-session B2B context: No persistence between sessions. Your CRM agent meets every prospect cold, every time. No deduplication or contradiction management. A prospect's budget moved from $80K to $120K across three sessions. Which number does your agent use in the fourth? No entity graph. Knowing Sarah is the engineering veto, she's at TechVision, and TechVision is expanding to APAC in Q3 ā€” those are relationships that need to be traversable, not flat stored text. Then the Q&A happened. A builder asked what nobody writes about in courses: what if the same user ID carries memories from two contexts that must never cross-contaminate ā€” a seller and a buyer on the same platform, or a patient and a caregiver in healthcare? One memory set poisoning the other. Rudraj's answer was honest: "That's an interesting edge case. There's no dynamic way to deal with this right now." And then he walked through how to think about scoping it manually ā€” namespace isolation at the Run ID level, context-specific extraction instructions, promoting shared signals deliberately. That answer is more useful than most content you'll find in any AI course. It's what happens when real builders push the edges of a real system live. Credit to Rudraj Mehta for the transparency, and to Aishwarya Srinivasan, Arvind Narayanamurthy for creating a format where that kind of question gets asked and answered in public. Full B2B walkthrough with an interactive example here: lnkd.in/gdHk4WNm
Is your AI actually remembering anything, or is it just repeating itself? In B2B work, 'memory' isn't just about recalling factsā€”itā€™s about preventing context from poisoning the next decision. #GenAcademy had Rudraj Mehta (Product Lead at #Mem0) walk through how memory algorithms actually work ā€” the session demo was a restaurant chatbot that remembers you like pasta and avoids mushrooms. Simple example, makes the architecture click. But the three phases he described ā€” extract, consolidate, retrieve ā€” are exactly what #B2BAI teams keep rebuilding from scratch when their agents lose context between sales calls. The things #RAG and Claude alone don't cover in a cross-session B2B context: No persistence between sessions. Your CRM agent meets every prospect cold, every time. No deduplication or contradiction management. A prospect's budget moved from $80K to $120K across three sessions. Which number does your agent use in the fourth? No entity graph. Knowing Sarah is the engineering veto, she's at TechVision, and TechVision is expanding to APAC in Q3 ā€” those are relationships that need to be traversable, not flat stored text. Then the Q&A happened. A builder asked what nobody writes about in courses: what if the same user ID carries memories from two contexts that must never cross-contaminate ā€” a seller and a buyer on the same platform, or a patient and a caregiver in healthcare? One memory set poisoning the other. Rudraj's answer was honest: "That's an interesting edge case. There's no dynamic way to deal with this right now." And then he walked through how to think about scoping it manually ā€” namespace isolation at the Run ID level, context-specific extraction instructions, promoting shared signals deliberately. That answer is more useful than most content you'll find in any AI course. It's what happens when real builders push the edges of a real system live. Credit to Rudraj Mehta for the transparency, and to Aishwarya Srinivasan, Arvind Narayanamurthy for creating a format where that kind of question gets asked and answered in public. Full B2B walkthrough with an interactive example here: lnkd.in/gdHk4WNm

Anna Ayvazyan
Week 3 : I built a multi-agent tax-research system that answers questions about current 401(k) and retirement rules . šŸ”¹ A different model for every agent A Claude Sonnet orchestrator decomposes the question. Fast, cheap Haiku researchers run search in parallel. GPT-4o-mini extracts structured claims. The result: frontier reasoning where it matters, cheap horsepower where it doesn't. šŸ”¹ Decorrelated errors as a feature. The verifier runs on a different model family than the extractor that produced the claims , so it catches mistakes a same-model check would happily rubber-stamp. The system actually enforces this at startup. šŸ”¹ Ground truth beats vibes. A deterministic knowledge base of curated IRS figures is the verifier's source of truth. Every claim carries a source URL, snippet, and date. The synthesizer can only cite supported, authoritative claims ,and it explains the rules without crossing into personalized financial advice. šŸ”¹ Built on LangGraph with parallel fan-out and a bounded orchestration loop that sends the graph back to re-plan when it spots a gap. The big lesson: trustworthy AI isn't about one giant model knowing everything. It's about architecture , routing the right model to the right job, and engineering in disagreement so errors surface instead of slipping through. Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy :) #AI #MultiAgent #LangGraph #LLM #MachineLearning #AIEngineering lnkd.in/giTt-TSs
Week 3 : I built a multi-agent tax-research system that answers questions about current 401(k) and retirement rules . šŸ”¹ A different model for every agent A Claude Sonnet orchestrator decomposes the question. Fast, cheap Haiku researchers run search in parallel. GPT-4o-mini extracts structured claims. The result: frontier reasoning where it matters, cheap horsepower where it doesn't. šŸ”¹ Decorrelated errors as a feature. The verifier runs on a different model family than the extractor that produced the claims , so it catches mistakes a same-model check would happily rubber-stamp. The system actually enforces this at startup. šŸ”¹ Ground truth beats vibes. A deterministic knowledge base of curated IRS figures is the verifier's source of truth. Every claim carries a source URL, snippet, and date. The synthesizer can only cite supported, authoritative claims ,and it explains the rules without crossing into personalized financial advice. šŸ”¹ Built on LangGraph with parallel fan-out and a bounded orchestration loop that sends the graph back to re-plan when it spots a gap. The big lesson: trustworthy AI isn't about one giant model knowing everything. It's about architecture , routing the right model to the right job, and engineering in disagreement so errors surface instead of slipping through. Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy :) #AI #MultiAgent #LangGraph #LLM #MachineLearning #AIEngineering lnkd.in/giTt-TSs

Lawin (Sunny) Bunga
šŸ“† Week 3 - The Gen Academy āš½ World Cup fever has taken over Toronto right now ā€” the city is absolutely electric. šŸ† So this week I built something that mixes the two things on my mind: the beautiful game, and agentic AI. Meet #VirtualSoccerArena ā€” #Week3 of the Agentic AI Bootcamp. It's an MCP server that lets you book a soccer pitch just by chatting with Claude. Ask "what's open June 28, what's the weather, and what's the rain policy?" ā€” and Claude pulls the forecast, checks live availability, answers the policy question from a 100-page rulebook via RAG, and books your slot. Four tools, one ReAct loop, one natural conversation. A web app then reads the same booking and unlocks a goalkeeper mini-game. šŸŽ® Playful on the outside. But under the hood, it's a masterclass in what production agentic AI demands: šŸ§  One brain, two front doors ā€” Claude and the web app share a single core and a single source of truth. Zero drift. šŸ”’ Human-in-the-loop where it counts ā€” Claude reasons and drafts on its own, but every irreversible action (booking, email) waits for a human yes. That's the line between a helpful agent and a liability. šŸ—„ļø The right tool for the data ā€” SQLite for transactional bookings, LlamaIndex + RAG for the policy PDF. Match the store to the shape, not the hype. The takeaway that'll stay with me: agentic AI is really a distributed-systems problem wearing a friendly chat interface. MCP gives you a clean integration contract, shared state keeps your front doors honest, RAG grounds the agent in real knowledge, and human-in-the-loop gates keep it safe ā€” but it all only holds together because the logic lives in one core and one source of truth. The model is the exciting part; the architecture is what earns the trust. And it scales beyond soccer ā€” the same Quote-to-Booking blueprint fits banking, telecom, healthcare, oil & gas, and e-commerce. Kudos to Aishwarya Srinivasan and Arvind Narayanamurthy for curating projects that don't just teach the theory, but force you to stress-test it. Building these systems from the ground up is the only way to truly internalize the concepts. šŸ‘ Game on. šŸ”— Code: lnkd.in/gKEeEKpN Slides below šŸ‘‡ #AgenticAI #MCP #AIArchitecture #WorldCup2026 #SolutionArchitecture #RAG
šŸ“† Week 3 - The Gen Academy āš½ World Cup fever has taken over Toronto right now ā€” the city is absolutely electric. šŸ† So this week I built something that mixes the two things on my mind: the beautiful game, and agentic AI. Meet #VirtualSoccerArena ā€” #Week3 of the Agentic AI Bootcamp. It's an MCP server that lets you book a soccer pitch just by chatting with Claude. Ask "what's open June 28, what's the weather, and what's the rain policy?" ā€” and Claude pulls the forecast, checks live availability, answers the policy question from a 100-page rulebook via RAG, and books your slot. Four tools, one ReAct loop, one natural conversation. A web app then reads the same booking and unlocks a goalkeeper mini-game. šŸŽ® Playful on the outside. But under the hood, it's a masterclass in what production agentic AI demands: šŸ§  One brain, two front doors ā€” Claude and the web app share a single core and a single source of truth. Zero drift. šŸ”’ Human-in-the-loop where it counts ā€” Claude reasons and drafts on its own, but every irreversible action (booking, email) waits for a human yes. That's the line between a helpful agent and a liability. šŸ—„ļø The right tool for the data ā€” SQLite for transactional bookings, LlamaIndex + RAG for the policy PDF. Match the store to the shape, not the hype. The takeaway that'll stay with me: agentic AI is really a distributed-systems problem wearing a friendly chat interface. MCP gives you a clean integration contract, shared state keeps your front doors honest, RAG grounds the agent in real knowledge, and human-in-the-loop gates keep it safe ā€” but it all only holds together because the logic lives in one core and one source of truth. The model is the exciting part; the architecture is what earns the trust. And it scales beyond soccer ā€” the same Quote-to-Booking blueprint fits banking, telecom, healthcare, oil & gas, and e-commerce. Kudos to Aishwarya Srinivasan and Arvind Narayanamurthy for curating projects that don't just teach the theory, but force you to stress-test it. Building these systems from the ground up is the only way to truly internalize the concepts. šŸ‘ Game on. šŸ”— Code: lnkd.in/gKEeEKpN Slides below šŸ‘‡ #AgenticAI #MCP #AIArchitecture #WorldCup2026 #SolutionArchitecture #RAG

Navneet Singh
MCP is getting more production-ready. The biggest shift Iā€™m watching is not ā€œmore tools.ā€ It is a better infrastructure for remote agent systems. The latest MCP spec keeps the familiar initialization flow, but adds important production pieces: - Streamable HTTP as the main remote transport - optional session management with MCP-Session-Id - polling-friendly SSE behavior - resumability with Last-Event-ID - stronger authorization discovery with OIDC support - experimental tasks for durable requests - improved elicitation for human-in-the-loop workflows The takeaway for AI engineers: MCP is moving from ā€œlocal connector demoā€ toward a real application layer for agentic systems. That matters because production agents need more than tool calls. They need: - secure authorization - resumable execution - clear user consent - human approval flows - durable task tracking - infrastructure that survives real network conditions I am learning about this in detail, thanks to the Cohort I am enrolled in with great mentors like Aishwarya Srinivasan and Arvind Narayanamurthy Iā€™m especially interested in how teams will manage state across remote MCP servers. Do you keep state in the server session? Push it into durable task storage? Or design the agent loop to be mostly stateless and resumable? This feels like one of the big architecture questions for agentic AI. #MCP #AgenticAI #AIEngineering #SystemArchitecture #CloudComputing #SoftwareEngineering #BuildInPublic
MCP is getting more production-ready. The biggest shift Iā€™m watching is not ā€œmore tools.ā€ It is a better infrastructure for remote agent systems. The latest MCP spec keeps the familiar initialization flow, but adds important production pieces: - Streamable HTTP as the main remote transport - optional session management with MCP-Session-Id - polling-friendly SSE behavior - resumability with Last-Event-ID - stronger authorization discovery with OIDC support - experimental tasks for durable requests - improved elicitation for human-in-the-loop workflows The takeaway for AI engineers: MCP is moving from ā€œlocal connector demoā€ toward a real application layer for agentic systems. That matters because production agents need more than tool calls. They need: - secure authorization - resumable execution - clear user consent - human approval flows - durable task tracking - infrastructure that survives real network conditions I am learning about this in detail, thanks to the Cohort I am enrolled in with great mentors like Aishwarya Srinivasan and Arvind Narayanamurthy Iā€™m especially interested in how teams will manage state across remote MCP servers. Do you keep state in the server session? Push it into durable task storage? Or design the agent loop to be mostly stateless and resumable? This feels like one of the big architecture questions for agentic AI. #MCP #AgenticAI #AIEngineering #SystemArchitecture #CloudComputing #SoftwareEngineering #BuildInPublic

Mimisha Mittal
šŸ’µ Currency conversion tells you what something costs. It doesn't tell you whether it's actually affordable. That's the problem I explored in my latest project: š†š„šØš›ššš„ š‚šØš¬š­ š€ššÆš¢š¬šØš«. The platform combines purchasing power parity, cost-of-living data, and AI-powered insights to help travelers, international students, digital nomads, and relocating professionals understand what prices actually mean across different economies. For example, a $6 coffee in San Francisco may seem expensive when converted to another currency, but is it actually expensive relative to local purchasing power? Building this project reinforced something I've been learning about AI: The value isn't in generating answers. It's in helping people make better decisions. GitHub: lnkd.in/gUASM2ny Thank you to Aishwarya Srinivasan and Arvind Narayanamurthy as I continue learning and building in AI, their content, guidance, and practical approach have been incredibly valuable in helping me. What other everyday decisions could AI help people make better? #AI #GenerativeAI #ProductManagement #BuildInPublic #ArtificialIntelligence
šŸ’µ Currency conversion tells you what something costs. It doesn't tell you whether it's actually affordable. That's the problem I explored in my latest project: š†š„šØš›ššš„ š‚šØš¬š­ š€ššÆš¢š¬šØš«. The platform combines purchasing power parity, cost-of-living data, and AI-powered insights to help travelers, international students, digital nomads, and relocating professionals understand what prices actually mean across different economies. For example, a $6 coffee in San Francisco may seem expensive when converted to another currency, but is it actually expensive relative to local purchasing power? Building this project reinforced something I've been learning about AI: The value isn't in generating answers. It's in helping people make better decisions. GitHub: lnkd.in/gUASM2ny Thank you to Aishwarya Srinivasan and Arvind Narayanamurthy as I continue learning and building in AI, their content, guidance, and practical approach have been incredibly valuable in helping me. What other everyday decisions could AI help people make better? #AI #GenerativeAI #ProductManagement #BuildInPublic #ArtificialIntelligence

Lakshmi Raghavan
The term "agentic AI" is getting stretched to cover three very different things. Worth untangling, especially if you're scoping your own build. This is something Arvind Narayanamurthy and Aishwarya Srinivasan touched on earlier in the cohort, and it's been resonating with me ever since. Take a simple use case: "help me stay under $500 for groceries and dining out this month." š—Ŗš—¼š—暝—øš—³š—¹š—¼š˜„ š—®š˜‚š˜š—¼š—ŗš—®š˜š—¶š—¼š—» runs a fixed sequence, tracks spending, and sends an alert when it crosses $500. No reasoning, no adaptation. Reliable as a Burglar alarm, but it only tells you after you've already blown the budget. š—„š—”š—š adds reasoning over retrieved context: it pulls your transaction history and spending patterns, and generates a recommendation grounded in both: "at this pace, you'll hit $500 by the 22nd." One pass ā†’ retrieve, reason, respond. Smarter, but it stops there. It told you the problem; it didn't do anything about it. š—”š—“š—²š—»š˜š—¶š—° š—”š—œ closes the loop. The system spots the overspend risk, decides on its own to dig into which subcategory is driving it, proposes pausing two unused subscriptions, checks if that's enough, and if it isn't ā†’ then proposes shifting funds from a category with slack, checks again, and only stops once it's actually found a path under $500 (or reports back honestly if it can't). That repeated act ā†’ check ā†’ act-again cycle, happening without you re-prompting at each step, is what makes it agentic. Remove that loop and leave just one shot, and it's not agentic anymore, just RAG with extra confidence. A natural follow-up: what if the sub-tasks ran in parallel instead of one after another? That's š—ŗš˜‚š—¹š˜š—¶-š—®š—“š—²š—»š˜: a spend-analysis agent, a subscription agent, and a reallocation agent each working their angle independently, with an orchestrator merging the results into one recommendation. Sounds more sophisticated. Often isn't worth it. Multi-agent earns its complexity when tasks are genuinely independent and expensive enough that parallelizing actually saves time. For one month's grocery budget, a single agent looping through the same checks sequentially gets you there just as well ā†’ faster, and with far less to debug. The coordination overhead of multiple agents (reconciling outputs, more API calls, more failure points) isn't free. The real test, regardless of architecture: if you removed the loop, would the system behave the same way? If yes, it was never agentic to begin with, just a single LLM call wearing the label. And the same test applies one level up: if you removed an agent, would the outcome change? That's the harder question I've been sitting with on my own work.
The term "agentic AI" is getting stretched to cover three very different things. Worth untangling, especially if you're scoping your own build. This is something Arvind Narayanamurthy and Aishwarya Srinivasan touched on earlier in the cohort, and it's been resonating with me ever since. Take a simple use case: "help me stay under $500 for groceries and dining out this month." š—Ŗš—¼š—暝—øš—³š—¹š—¼š˜„ š—®š˜‚š˜š—¼š—ŗš—®š˜š—¶š—¼š—» runs a fixed sequence, tracks spending, and sends an alert when it crosses $500. No reasoning, no adaptation. Reliable as a Burglar alarm, but it only tells you after you've already blown the budget. š—„š—”š—š adds reasoning over retrieved context: it pulls your transaction history and spending patterns, and generates a recommendation grounded in both: "at this pace, you'll hit $500 by the 22nd." One pass ā†’ retrieve, reason, respond. Smarter, but it stops there. It told you the problem; it didn't do anything about it. š—”š—“š—²š—»š˜š—¶š—° š—”š—œ closes the loop. The system spots the overspend risk, decides on its own to dig into which subcategory is driving it, proposes pausing two unused subscriptions, checks if that's enough, and if it isn't ā†’ then proposes shifting funds from a category with slack, checks again, and only stops once it's actually found a path under $500 (or reports back honestly if it can't). That repeated act ā†’ check ā†’ act-again cycle, happening without you re-prompting at each step, is what makes it agentic. Remove that loop and leave just one shot, and it's not agentic anymore, just RAG with extra confidence. A natural follow-up: what if the sub-tasks ran in parallel instead of one after another? That's š—ŗš˜‚š—¹š˜š—¶-š—®š—“š—²š—»š˜: a spend-analysis agent, a subscription agent, and a reallocation agent each working their angle independently, with an orchestrator merging the results into one recommendation. Sounds more sophisticated. Often isn't worth it. Multi-agent earns its complexity when tasks are genuinely independent and expensive enough that parallelizing actually saves time. For one month's grocery budget, a single agent looping through the same checks sequentially gets you there just as well ā†’ faster, and with far less to debug. The coordination overhead of multiple agents (reconciling outputs, more API calls, more failure points) isn't free. The real test, regardless of architecture: if you removed the loop, would the system behave the same way? If yes, it was never agentic to begin with, just a single LLM call wearing the label. And the same test applies one level up: if you removed an agent, would the outcome change? That's the harder question I've been sitting with on my own work.

The Gen Academy
Week 2 of our Mastering Agentic AI Certification, May Cohort is officially a wrap! Huge congratulations to Santosh, our AI Quiz Wiz of the Week! Arvind and Aishwarya are incredibly proud of what they're building with The Gen Academy. The engagement, curiosity, and energy from this cohort has been something truly special to witness. One thing they care deeply about is live learning. Learning AI is not just about consuming information. It is about asking questions, challenging ideas, learning alongside others, and building intuition through discussion. The energy that comes from a live classroom is something recorded lessons simply cannot replicate. That is why every session includes live Q&A, open discussions, and weekly quizzes that make learning interactive and fun. This week, Santosh earned the title of AI Quiz Wiz of the Week. Well deserved! P.S. See that "Artificially Intelligent" hat at the top? Santosh is officially getting one. They take quiz victories very seriously around here. Want to be part of a cohort like this? The August Cohort for the Mastering Agentic AI Certification is now open. Live sessions, real discussions, weekly quizzes, and a community that actually shows up for each other. šŸ‘‡ Link in the comments ā€” come join us.
Week 2 of our Mastering Agentic AI Certification, May Cohort is officially a wrap! Huge congratulations to Santosh, our AI Quiz Wiz of the Week! Arvind and Aishwarya are incredibly proud of what they're building with The Gen Academy. The engagement, curiosity, and energy from this cohort has been something truly special to witness. One thing they care deeply about is live learning. Learning AI is not just about consuming information. It is about asking questions, challenging ideas, learning alongside others, and building intuition through discussion. The energy that comes from a live classroom is something recorded lessons simply cannot replicate. That is why every session includes live Q&A, open discussions, and weekly quizzes that make learning interactive and fun. This week, Santosh earned the title of AI Quiz Wiz of the Week. Well deserved! P.S. See that "Artificially Intelligent" hat at the top? Santosh is officially getting one. They take quiz victories very seriously around here. Want to be part of a cohort like this? The August Cohort for the Mastering Agentic AI Certification is now open. Live sessions, real discussions, weekly quizzes, and a community that actually shows up for each other. šŸ‘‡ Link in the comments ā€” come join us.

Navneet Singh
Attended a guest lecture by Arjun Patel, Senior Developer Advocate at Pinecone, on retrieval patterns for RAG pipelines. My head is full of ideas. Here are my biggest takeaways: 1. How you store your data = how you can query it This sounds obvious until you actually build a RAG system. The embedding model you choose, how you chunk your data, and which index structure you use all shape the kinds of retrieval you can support later. Make those decisions intentionally. 2. Retrieval is not one thing Arjun walked through a live bird search demo covering: - Dense semantic search for vocabulary mismatch - Full-text / keyword search for precision queries and rare terms - Visual/image search using shared text-image vector space - Hybrid + filter for compound queries - Coarse-to-fine retrieval by filtering first, then searching inside the narrowed context - Agentic search, where Claude decides which tool to call based on query type 3. Evals do not have to be scary You do not need 1,000 labeled examples to start. Start with 5 to 6 queries you care about. Define what a good result looks like. Check whether your retrieval is actually improving. Diagnose first. Scale later. The mental model I am taking away: Improving retrieval improves RAG. Everything else is downstream of this. I am already applying these ideas to CohortBrain, a GraphRAG + Vector RAG knowledge system I am building for my bootcamp. The timing of this lecture could not have been better. If you are building RAG systems and not thinking deeply about your retrieval layer, this is your sign. Thanks to Arjun Patel and Aishwarya Srinivasan , Arvind Narayanamurthy for making this happen. #RAG #VectorSearch #Pinecone #LangChain #AIEngineering #BuildInPublic #GenerativeAI
Attended a guest lecture by Arjun Patel, Senior Developer Advocate at Pinecone, on retrieval patterns for RAG pipelines. My head is full of ideas. Here are my biggest takeaways: 1. How you store your data = how you can query it This sounds obvious until you actually build a RAG system. The embedding model you choose, how you chunk your data, and which index structure you use all shape the kinds of retrieval you can support later. Make those decisions intentionally. 2. Retrieval is not one thing Arjun walked through a live bird search demo covering: - Dense semantic search for vocabulary mismatch - Full-text / keyword search for precision queries and rare terms - Visual/image search using shared text-image vector space - Hybrid + filter for compound queries - Coarse-to-fine retrieval by filtering first, then searching inside the narrowed context - Agentic search, where Claude decides which tool to call based on query type 3. Evals do not have to be scary You do not need 1,000 labeled examples to start. Start with 5 to 6 queries you care about. Define what a good result looks like. Check whether your retrieval is actually improving. Diagnose first. Scale later. The mental model I am taking away: Improving retrieval improves RAG. Everything else is downstream of this. I am already applying these ideas to CohortBrain, a GraphRAG + Vector RAG knowledge system I am building for my bootcamp. The timing of this lecture could not have been better. If you are building RAG systems and not thinking deeply about your retrieval layer, this is your sign. Thanks to Arjun Patel and Aishwarya Srinivasan , Arvind Narayanamurthy for making this happen. #RAG #VectorSearch #Pinecone #LangChain #AIEngineering #BuildInPublic #GenerativeAI

Lakshmi Raghavan
For Week 2 of my Agentic AI cohort, I built a Rare Disease Workup Assistant ā€” an š€šˆ-š©šØš°šžš«šžš šœš„š¢š§š¢šœššš„ ššžšœš¢š¬š¢šØš§ š¬š®š©š©šØš«š­ š­šØšØš„ that helps primary care physicians identify potential rare diseases from free-text patient descriptions in under 30 seconds. š“š”šž š©š«šØš›š„šžš¦ š°šØš«š­š” š¬šØš„šÆš¢š§š  300 million people worldwide live with a rare disease. The average diagnostic odyssey takes 4ā€“6 years of misdiagnoses, unnecessary treatments, and unanswered questions for patients and families. Primary care physicians are the first door most of them walk through. Expected to recognize patterns across 7,000+ conditions, with 15-minute appointment slots and limited specialist access. That gap ā€” between what physicians are asked to do and what's humanly possible ā€” is where AI can make a difference. Not by replacing clinical judgment. By making sure the right question gets asked before the patient leaves the room. š–š”ššš­ šˆ š›š®š¢š„š­ A multi-stage Hybrid RAG pipeline ā€” agentic phenotype expansion, BM25 + vector retrieval via Qdrant, intelligent reranking, and grounded generation via Claude Sonnet. All orchestrated through n8n. š“š”šž šžšÆššš„ š«šžš¬š®š„š­š¬ āœ… 95.7% Top-1 retrieval accuracy āœ… 83% Faithfulness āœ… 95% Answer Relevance Strong numbers. But the most instructive result wasn't one of these. The result that made me stop and think 71% š¬š©š®š«š¢šØš®š¬ šœšØš§šŸš¢ššžš§šœšž š«ššš­šž on negative controls ā€” the system flagged a potential rare disease 5 out of 7 times when presented with common condition descriptions. First instinct: that's a problem. Second instinct ā€” after sitting with the confusion matrix: it isn't. A false negative in rare disease workup means a missed diagnosis. Another year on the diagnostic odyssey. A patient who walked in with answers walking out without them. A false positive means a physician follows an extra diagnostic thread. Inconvenient. Not harmful. A system calibrated toward sensitivity "one that would rather flag too much than miss something real" is not a flawed system. It's a responsibly designed one. The next iteration needs a confidence calibration layer ā€” a mechanism that can distinguish "this genuinely looks rare" from "I'm uncertain, so I'm flagging it." That's the next build. š–š”ššš­ š­š”š¢š¬ š©š«šØš£šžšœš­ š­ššš®š š”š­ š¦šž Getting a RAG system to retrieve accurately is a technical challenge. Getting one that a physician will actually trust ā€” transparent sourcing, calibrated confidence, explainable outputs ā€” is a design challenge. The eval numbers open the door. The thinking behind them is what builds the trust. Grateful to Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen Academy for building a program that measures you not by what you learned, but by what you built and why. As Aishwarya Srinivasan says ā€” "It's a marathon!" Week 3 starts now. šŸƒā€ā™€ļø #GenAI #RAG #HealthcareAI #RareDiseases #ClinicalAI #AIEvaluation #MasteringAI #TheGenAcademy
For Week 2 of my Agentic AI cohort, I built a Rare Disease Workup Assistant ā€” an š€šˆ-š©šØš°šžš«šžš šœš„š¢š§š¢šœššš„ ššžšœš¢š¬š¢šØš§ š¬š®š©š©šØš«š­ š­šØšØš„ that helps primary care physicians identify potential rare diseases from free-text patient descriptions in under 30 seconds. š“š”šž š©š«šØš›š„šžš¦ š°šØš«š­š” š¬šØš„šÆš¢š§š  300 million people worldwide live with a rare disease. The average diagnostic odyssey takes 4ā€“6 years of misdiagnoses, unnecessary treatments, and unanswered questions for patients and families. Primary care physicians are the first door most of them walk through. Expected to recognize patterns across 7,000+ conditions, with 15-minute appointment slots and limited specialist access. That gap ā€” between what physicians are asked to do and what's humanly possible ā€” is where AI can make a difference. Not by replacing clinical judgment. By making sure the right question gets asked before the patient leaves the room. š–š”ššš­ šˆ š›š®š¢š„š­ A multi-stage Hybrid RAG pipeline ā€” agentic phenotype expansion, BM25 + vector retrieval via Qdrant, intelligent reranking, and grounded generation via Claude Sonnet. All orchestrated through n8n. š“š”šž šžšÆššš„ š«šžš¬š®š„š­š¬ āœ… 95.7% Top-1 retrieval accuracy āœ… 83% Faithfulness āœ… 95% Answer Relevance Strong numbers. But the most instructive result wasn't one of these. The result that made me stop and think 71% š¬š©š®š«š¢šØš®š¬ šœšØš§šŸš¢ššžš§šœšž š«ššš­šž on negative controls ā€” the system flagged a potential rare disease 5 out of 7 times when presented with common condition descriptions. First instinct: that's a problem. Second instinct ā€” after sitting with the confusion matrix: it isn't. A false negative in rare disease workup means a missed diagnosis. Another year on the diagnostic odyssey. A patient who walked in with answers walking out without them. A false positive means a physician follows an extra diagnostic thread. Inconvenient. Not harmful. A system calibrated toward sensitivity "one that would rather flag too much than miss something real" is not a flawed system. It's a responsibly designed one. The next iteration needs a confidence calibration layer ā€” a mechanism that can distinguish "this genuinely looks rare" from "I'm uncertain, so I'm flagging it." That's the next build. š–š”ššš­ š­š”š¢š¬ š©š«šØš£šžšœš­ š­ššš®š š”š­ š¦šž Getting a RAG system to retrieve accurately is a technical challenge. Getting one that a physician will actually trust ā€” transparent sourcing, calibrated confidence, explainable outputs ā€” is a design challenge. The eval numbers open the door. The thinking behind them is what builds the trust. Grateful to Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen Academy for building a program that measures you not by what you learned, but by what you built and why. As Aishwarya Srinivasan says ā€” "It's a marathon!" Week 3 starts now. šŸƒā€ā™€ļø #GenAI #RAG #HealthcareAI #RareDiseases #ClinicalAI #AIEvaluation #MasteringAI #TheGenAcademy

Navneet Singh
Another week, another guest lecture that changed how I think about building AI systems. This time, Logan Markewich , Head of Open Source at LlamaIndex, walked us through document parsing, retrieval design, and a live financial due diligence agent. Here is what stuck with me: 1. PDFs are a mess, and that is the whole problem PDFs were not built to be machine-readable. Data is scattered. Reading order is broken. Tables are chaos. LlamaIndex built both LlamaParse and their document parsing stack to help solve this: - fast local parsing - managed parsing APIs - chart understanding - semantic formatting - granular bounding boxes The better the parse, the better the downstream AI system. 2. You probably do not need embeddings yet This one hit different. Logan built a financial RAG agent with zero vector embeddings. Just keyword search over parsed documents. His take: For 6ā€“50 docs, keyword search plus dynamic context windows can beat embeddings. Embeddings are for narrowing search space at scale. They are not always the best tool for precision. 3. The citation problem is harder than it looks The agent cited sources with bounding box highlights on the actual PDF page. Simple idea. Hard execution. Models subtly reword things. "$394,000" can become "394 billion." Now your exact match breaks. Loganā€™s solution was a waterfall of fallbacks: exact match ā†’ whitespace stripped ā†’ currency stripped ā†’ longest subtoken match And if all else fails, still surface the page number. Better than pretending the citation is perfect. 4. ParseBench: build your own benchmark LlamaIndex built their own eval set because existing academic benchmarks did not reflect what customers actually struggled with. They evaluated across: - table accuracy - content faithfulness - visual grounding - chart understanding - semantic formatting Sound familiar? It is the same mindset as retrieval evals: Start with the problems you care about, not only the benchmarks that already exist. I am using this thinking to sharpen the evaluation layer in CohortBrain, my GraphRAG + Vector RAG system for the bootcamp. The timing could not be better. Thanks to Logan Markewich and Aishwarya Srinivasan, Arvind Narayanamurthy for making these sessions happen. #LlamaIndex #RAG #DocumentParsing #AIEngineering #BuildInPublic #GenerativeAI #LLM
Another week, another guest lecture that changed how I think about building AI systems. This time, Logan Markewich , Head of Open Source at LlamaIndex, walked us through document parsing, retrieval design, and a live financial due diligence agent. Here is what stuck with me: 1. PDFs are a mess, and that is the whole problem PDFs were not built to be machine-readable. Data is scattered. Reading order is broken. Tables are chaos. LlamaIndex built both LlamaParse and their document parsing stack to help solve this: - fast local parsing - managed parsing APIs - chart understanding - semantic formatting - granular bounding boxes The better the parse, the better the downstream AI system. 2. You probably do not need embeddings yet This one hit different. Logan built a financial RAG agent with zero vector embeddings. Just keyword search over parsed documents. His take: For 6ā€“50 docs, keyword search plus dynamic context windows can beat embeddings. Embeddings are for narrowing search space at scale. They are not always the best tool for precision. 3. The citation problem is harder than it looks The agent cited sources with bounding box highlights on the actual PDF page. Simple idea. Hard execution. Models subtly reword things. "$394,000" can become "394 billion." Now your exact match breaks. Loganā€™s solution was a waterfall of fallbacks: exact match ā†’ whitespace stripped ā†’ currency stripped ā†’ longest subtoken match And if all else fails, still surface the page number. Better than pretending the citation is perfect. 4. ParseBench: build your own benchmark LlamaIndex built their own eval set because existing academic benchmarks did not reflect what customers actually struggled with. They evaluated across: - table accuracy - content faithfulness - visual grounding - chart understanding - semantic formatting Sound familiar? It is the same mindset as retrieval evals: Start with the problems you care about, not only the benchmarks that already exist. I am using this thinking to sharpen the evaluation layer in CohortBrain, my GraphRAG + Vector RAG system for the bootcamp. The timing could not be better. Thanks to Logan Markewich and Aishwarya Srinivasan, Arvind Narayanamurthy for making these sessions happen. #LlamaIndex #RAG #DocumentParsing #AIEngineering #BuildInPublic #GenerativeAI #LLM

Vyas S
šŸš€ The Future of AI Isn't Bigger Models. It's AI Daemons. Over the past few weeks, I've been diving deep into research on AI agents, memory architectures, autonomous systems, and the emerging concept of AI Daemons. One thing has become increasingly clear: The future of AI is not just about building larger models. It's about creating persistent, memory-rich, autonomous systems that can continuously observe, reason, learn, and act. In traditional computing, a daemon is a background process that runs continuously without direct user interaction. In modern AI, this idea is evolving into intelligent agents capable of maintaining context, managing memory, planning actions, coordinating with other agents, and executing complex workflows autonomously. Recent research highlights several key developments: šŸ”¹ MemGPT introduces the concept of treating Large Language Models like operating systems with hierarchical memory management. šŸ”¹ Agentic AI architectures are moving toward structured systems consisting of perception, reasoning, planning, tool usage, collaboration, and execution layers. šŸ”¹ Persistent memory frameworks such as Mem0 and soul.py are exploring how AI agents can maintain identity, continuity, and long-term memory across sessions. šŸ”¹ Multi-agent frameworks including LangChain, LlamaIndex, CrewAI, and Swarm are enabling specialized agents to collaborate on increasingly sophisticated tasks. šŸ”¹ Memory OS research proposes that memory itself should function as an independent operating layer for AI systems, separate from the reasoning model. The emerging architecture increasingly resembles: Perception ā†’ Memory ā†’ Planning ā†’ Action ā†’ Audit Instead of simply responding to prompts, future AI systems may continuously monitor environments, accumulate knowledge, adapt from experience, and execute decisions through coordinated networks of autonomous agents. What excites me most is that we're witnessing a shift from AI as a tool to AI as infrastructure. The academic and engineering communities appear to be converging on a common direction: The future of AI is not smarter standalone models. It is persistent, memory-driven, multi-agent daemon systems operating autonomously at scale. A special thanks to Yannic Kilcher for his exceptional research paper breakdowns and technical analyses. His content has been incredibly valuable in helping me understand complex AI concepts, agent architectures, memory systems, and emerging research directions. If you're interested in understanding AI research beyond surface-level tutorials, I highly recommend exploring his work. Also grateful to Aishwarya Srinivasan for her practical AI engineering content, which has helped bridge the gap between research papers and real-world implementation. As someone actively building AI projects and exploring agentic systems, this research has fundamentally changed how I think about the next generation of intelligent applications. What are your thoughts? Will AI Daemons become the foundation of future software systems? #ArtificialIntelligence #AIAgents #AgenticAI #GenerativeAI #MachineLearning #LLM #MemGPT #RAG #MemorySystems #MultiAgentSystems #AIEngineering #DataScience #FutureOfAI #AIResearch #CrewAI #LangChain #LlamaIndex #Innovation
šŸš€ The Future of AI Isn't Bigger Models. It's AI Daemons. Over the past few weeks, I've been diving deep into research on AI agents, memory architectures, autonomous systems, and the emerging concept of AI Daemons. One thing has become increasingly clear: The future of AI is not just about building larger models. It's about creating persistent, memory-rich, autonomous systems that can continuously observe, reason, learn, and act. In traditional computing, a daemon is a background process that runs continuously without direct user interaction. In modern AI, this idea is evolving into intelligent agents capable of maintaining context, managing memory, planning actions, coordinating with other agents, and executing complex workflows autonomously. Recent research highlights several key developments: šŸ”¹ MemGPT introduces the concept of treating Large Language Models like operating systems with hierarchical memory management. šŸ”¹ Agentic AI architectures are moving toward structured systems consisting of perception, reasoning, planning, tool usage, collaboration, and execution layers. šŸ”¹ Persistent memory frameworks such as Mem0 and soul.py are exploring how AI agents can maintain identity, continuity, and long-term memory across sessions. šŸ”¹ Multi-agent frameworks including LangChain, LlamaIndex, CrewAI, and Swarm are enabling specialized agents to collaborate on increasingly sophisticated tasks. šŸ”¹ Memory OS research proposes that memory itself should function as an independent operating layer for AI systems, separate from the reasoning model. The emerging architecture increasingly resembles: Perception ā†’ Memory ā†’ Planning ā†’ Action ā†’ Audit Instead of simply responding to prompts, future AI systems may continuously monitor environments, accumulate knowledge, adapt from experience, and execute decisions through coordinated networks of autonomous agents. What excites me most is that we're witnessing a shift from AI as a tool to AI as infrastructure. The academic and engineering communities appear to be converging on a common direction: The future of AI is not smarter standalone models. It is persistent, memory-driven, multi-agent daemon systems operating autonomously at scale. A special thanks to Yannic Kilcher for his exceptional research paper breakdowns and technical analyses. His content has been incredibly valuable in helping me understand complex AI concepts, agent architectures, memory systems, and emerging research directions. If you're interested in understanding AI research beyond surface-level tutorials, I highly recommend exploring his work. Also grateful to Aishwarya Srinivasan for her practical AI engineering content, which has helped bridge the gap between research papers and real-world implementation. As someone actively building AI projects and exploring agentic systems, this research has fundamentally changed how I think about the next generation of intelligent applications. What are your thoughts? Will AI Daemons become the foundation of future software systems? #ArtificialIntelligence #AIAgents #AgenticAI #GenerativeAI #MachineLearning #LLM #MemGPT #RAG #MemorySystems #MultiAgentSystems #AIEngineering #DataScience #FutureOfAI #AIResearch #CrewAI #LangChain #LlamaIndex #Innovation

Sneha Upadhyayula
When you deeply understand a system, you stop explaining it through terminology and start explaining it through intuition. Thatā€™s exactly what I saw Chris Alexiuk ā€œThe Wizā€ from NVIDIA do while walking through a very complex RLVR training pipeline. Instead of starting with reward shaping, delayed credit assignment, multi-environment RL, reward hacking, Chris explained the system through things we humans already intuitively understand: 1. Models ā€œwant cookiesā€ ā†’ reward optimization 2. SFT ā†’ like learning arithmetic before solving harder problems 3. SWE-focused RLVR ā†’ like project grading instead of pop quizzes because feedback arrives much later 4. Reward hacking ā†’ a robot trained to find yellow cheese starts chasing bananas instead What made these analogies powerful wasnā€™t that they simplified the system away. They preserved the actual dynamics: incentives, feedback loops, delayed rewards, curriculum learning, proxy optimization, and generalization pressure. Thatā€™s what strong technical intuition feels like. You stop describing the mechanics alone and start seeing the underlying adaptive system. Personally, I think this lesson extends beyond AI. Every system eventually learns the proxy you reward. Models do. Organizations do. People do. Every adaptive system eventually learns the proxy it is rewarded on.... Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for facilitating this awesome session! #AI #Nemotron #Nvidia
When you deeply understand a system, you stop explaining it through terminology and start explaining it through intuition. Thatā€™s exactly what I saw Chris Alexiuk ā€œThe Wizā€ from NVIDIA do while walking through a very complex RLVR training pipeline. Instead of starting with reward shaping, delayed credit assignment, multi-environment RL, reward hacking, Chris explained the system through things we humans already intuitively understand: 1. Models ā€œwant cookiesā€ ā†’ reward optimization 2. SFT ā†’ like learning arithmetic before solving harder problems 3. SWE-focused RLVR ā†’ like project grading instead of pop quizzes because feedback arrives much later 4. Reward hacking ā†’ a robot trained to find yellow cheese starts chasing bananas instead What made these analogies powerful wasnā€™t that they simplified the system away. They preserved the actual dynamics: incentives, feedback loops, delayed rewards, curriculum learning, proxy optimization, and generalization pressure. Thatā€™s what strong technical intuition feels like. You stop describing the mechanics alone and start seeing the underlying adaptive system. Personally, I think this lesson extends beyond AI. Every system eventually learns the proxy you reward. Models do. Organizations do. People do. Every adaptive system eventually learns the proxy it is rewarded on.... Thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for facilitating this awesome session! #AI #Nemotron #Nvidia

Sham Sundar Hassan Chikkegowda
My take on a #GenAcademy guest lecture I attended today. First watch: lost me halfway through. Left thinking I just wasn't ready for the content. Second watch: started making sense. Third watch: 7 ideas I'd never seen in any course. The RCA on why I was slow: I was still thinking in traditional LLM mode. The lecture starts past that. If you're in the same spot, the blog I built links to my older posts on classical RL and LLM fundamentals ā€” read those first, then come back. The 7 takeaways aren't academic. They're the kind of thing that only surfaces once you're actually building with LLMs: ā€” Why RL needs SFT as a warm-up (and how little is enough) ā€” How models learn shortcuts instead of goals ā€” Why running 70+ environments at once isn't overhead ā€” it's the fix ā€” Why RLHF trained first produces a politely useless model Each concept has an interactive sandbox. lnkd.in/g22eJ-vv Thanks again to Chris Alexiuk for the lecture. Getting there took a few attempts, but it paid off. -------------------------------------- The irony isn't lost on me. Traditional LLM me rewatched the same lecture three times expecting different output ā€” which is literally single-environment reward hacking. Same shortcut, same result, no progress. The fix, as I now know, was to change the environment: stop watching and start building. That's when the signal kicked in. Next session I'm showing up with better priors. SFT complete. Echoing what Aishwarya Srinivasan and Arvind Narayanamurthy mentioned about handling the complexityā€”it was tough, but making the mental model is what matters.
My take on a #GenAcademy guest lecture I attended today. First watch: lost me halfway through. Left thinking I just wasn't ready for the content. Second watch: started making sense. Third watch: 7 ideas I'd never seen in any course. The RCA on why I was slow: I was still thinking in traditional LLM mode. The lecture starts past that. If you're in the same spot, the blog I built links to my older posts on classical RL and LLM fundamentals ā€” read those first, then come back. The 7 takeaways aren't academic. They're the kind of thing that only surfaces once you're actually building with LLMs: ā€” Why RL needs SFT as a warm-up (and how little is enough) ā€” How models learn shortcuts instead of goals ā€” Why running 70+ environments at once isn't overhead ā€” it's the fix ā€” Why RLHF trained first produces a politely useless model Each concept has an interactive sandbox. lnkd.in/g22eJ-vv Thanks again to Chris Alexiuk for the lecture. Getting there took a few attempts, but it paid off. -------------------------------------- The irony isn't lost on me. Traditional LLM me rewatched the same lecture three times expecting different output ā€” which is literally single-environment reward hacking. Same shortcut, same result, no progress. The fix, as I now know, was to change the environment: stop watching and start building. That's when the signal kicked in. Next session I'm showing up with better priors. SFT complete. Echoing what Aishwarya Srinivasan and Arvind Narayanamurthy mentioned about handling the complexityā€”it was tough, but making the mental model is what matters.

Swati Aggarwal
I was going through this slide in the Gen AI Building Blocks Bootcamp last week. 2026 Model Landscape. Open Source. Open Weights. Closed Source. And I realised ā€” I had been using these terms loosely for years. Treating "open source" and "open weights" as if they meant the same thing. They don't. And if you're building anything that touches sensitive data, the difference is not academic. Let me explain it the way it finally clicked for me. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Think of an AI model as a cake. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” The recipe = training code. Architecture, data, process ā€” everything that created the model. The finished cake = the weights. Billions of numbers that ARE the model's intelligence. When it gives you a smart answer ā€” those numbers are doing the work. Open Source ā†’ recipe AND cake. Bake your own from scratch. Full control. Examples: Llama 3.1, Mistral, Gemma 2. Open Weights ā†’ finished cake only. You can fine-tune it, run it, add your own flavours ā€” but can't recreate it from scratch. Examples: Meta Llama 3.1 instruct, Cohere Command R+, Snowflake Arctic. Closed Source ā†’ you taste the cake through a window. Never hold it. Never see the recipe. Pay for every slice. Examples: GPT-4o, Claude, Gemini. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” What this means when you're building ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Closed source ā€” fastest to start. Best reasoning. Just call the API. Perfect for prototyping and apps where privacy isn't a constraint. Open weights ā€” right when data cannot leave your infrastructure. Medical records. Financial statements. PII. Run it on your server, data never touches a third-party API. Open source ā€” for teams going deep. Reproduce training. Study architecture. Build derivatives. Essential for research. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” The mistake I see most often ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Teams assume "open" means private. It doesn't automatically. If you're calling someone else's hosted open-weights API ā€” your data is still leaving your machine. The privacy benefit only kicks in when you self-host. Privacy isn't something you add later. It's a decision that shapes your architecture from day one. The cake is only yours if you bake it in your own kitchen. Next post ā€” how to actually choose between these three when designing a real system. Three questions that make the decision obvious. Slide credit: Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen AI Academy ā€” Gen AI Building Blocks Bootcamp Week 1. #GenerativeAI #LLM #OpenSource #AIEngineering #PrivacyByDesign #HealthcareAI #MachineLearning #AIAgents #CloudSecurity #BuildingInPublic #AIBootcamp #ResponsibleAI
I was going through this slide in the Gen AI Building Blocks Bootcamp last week. 2026 Model Landscape. Open Source. Open Weights. Closed Source. And I realised ā€” I had been using these terms loosely for years. Treating "open source" and "open weights" as if they meant the same thing. They don't. And if you're building anything that touches sensitive data, the difference is not academic. Let me explain it the way it finally clicked for me. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Think of an AI model as a cake. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” The recipe = training code. Architecture, data, process ā€” everything that created the model. The finished cake = the weights. Billions of numbers that ARE the model's intelligence. When it gives you a smart answer ā€” those numbers are doing the work. Open Source ā†’ recipe AND cake. Bake your own from scratch. Full control. Examples: Llama 3.1, Mistral, Gemma 2. Open Weights ā†’ finished cake only. You can fine-tune it, run it, add your own flavours ā€” but can't recreate it from scratch. Examples: Meta Llama 3.1 instruct, Cohere Command R+, Snowflake Arctic. Closed Source ā†’ you taste the cake through a window. Never hold it. Never see the recipe. Pay for every slice. Examples: GPT-4o, Claude, Gemini. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” What this means when you're building ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Closed source ā€” fastest to start. Best reasoning. Just call the API. Perfect for prototyping and apps where privacy isn't a constraint. Open weights ā€” right when data cannot leave your infrastructure. Medical records. Financial statements. PII. Run it on your server, data never touches a third-party API. Open source ā€” for teams going deep. Reproduce training. Study architecture. Build derivatives. Essential for research. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” The mistake I see most often ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Teams assume "open" means private. It doesn't automatically. If you're calling someone else's hosted open-weights API ā€” your data is still leaving your machine. The privacy benefit only kicks in when you self-host. Privacy isn't something you add later. It's a decision that shapes your architecture from day one. The cake is only yours if you bake it in your own kitchen. Next post ā€” how to actually choose between these three when designing a real system. Three questions that make the decision obvious. Slide credit: Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen AI Academy ā€” Gen AI Building Blocks Bootcamp Week 1. #GenerativeAI #LLM #OpenSource #AIEngineering #PrivacyByDesign #HealthcareAI #MachineLearning #AIAgents #CloudSecurity #BuildingInPublic #AIBootcamp #ResponsibleAI

Ravi Kiran Uppalapati
As part of the Mastering Agentic AI cohort led by Aishwarya Srinivasan and Arvind Narayanamurthy, there's been constant encouragement to move beyond learning concepts and actually build something useful. This is a small outcome of that push. I built a UK FTTP Opportunity Map using public Ofcom and ONS data. The dashboard helps explore: 1.Where full-fibre availability is strong 2.Where gaps still remain 3.Which local authorities have the most premises without FTTP 4.Oppurtunity score using coverage levels and premises without FTTP More than the dashboard itself, the project helped me think practically about how Agentic AI can support real analytical workflows ā€” not just generating ideas, but shaping methodology, building the app, validating assumptions, documenting calculations, and getting it deployment-ready. Still early, still learning, but genuinely enjoying the process of building and experimenting. Grateful for the nudge to learn by doing. šŸ™ šŸ”— Live app: lnkd.in/eSnXc75M šŸ’» GitHub: lnkd.in/eTwVVbQu
As part of the Mastering Agentic AI cohort led by Aishwarya Srinivasan and Arvind Narayanamurthy, there's been constant encouragement to move beyond learning concepts and actually build something useful. This is a small outcome of that push. I built a UK FTTP Opportunity Map using public Ofcom and ONS data. The dashboard helps explore: 1.Where full-fibre availability is strong 2.Where gaps still remain 3.Which local authorities have the most premises without FTTP 4.Oppurtunity score using coverage levels and premises without FTTP More than the dashboard itself, the project helped me think practically about how Agentic AI can support real analytical workflows ā€” not just generating ideas, but shaping methodology, building the app, validating assumptions, documenting calculations, and getting it deployment-ready. Still early, still learning, but genuinely enjoying the process of building and experimenting. Grateful for the nudge to learn by doing. šŸ™ šŸ”— Live app: lnkd.in/eSnXc75M šŸ’» GitHub: lnkd.in/eTwVVbQu

Mansoor Mohammed
It's been close to a decade since I actually did some hands-on work in VS Code and GitHub. Courtesy of the Mastering Agentic AI course at The Gen Academy, I shipped my first vibe coded application an AI-Powered Product Description Generator (keeping things very simple!): a tool that helps eCommerce sellers generate fully optimized product listings. Here's what it does: šŸ“ Takes a product name, key features, an uploaded image, and a spec document as inputs šŸ›ļø Generates platform-specific listings for Amazon, Etsy, Shopify, and more šŸ” Produces SEO keywords ā€” primary, secondary, and long-tail šŸ¤– Also generates some AEO and GEO content All powered by GPT-4o, built in React, and live on GitHub Pages. šŸ‘‰ Try it here: lnkd.in/gP2bntEh What I learned building this: āœ… How to design and prompt multimodal AI API calls (text + image + document in a single request) āœ… How to structure prompts to enforce strict JSON output from LLMs āœ… How to go from an idea ā†’ a working prompt ā†’ a shipped app using vibe coding āœ… Version control and deployment via GitHub, GitHub Pages and Vercel A huge thank you to Aishwarya and Arvind for building a course that actually makes you ship things. :) I am in Week-1, so just getting started. More builds coming. šŸ”Ø #AgenticAI #GenerativeAI #VibeCode #OpenAI #eCommerce #GenAcademy #MasteringAgenticAI #AIProducts #LearningInPublic
It's been close to a decade since I actually did some hands-on work in VS Code and GitHub. Courtesy of the Mastering Agentic AI course at The Gen Academy, I shipped my first vibe coded application an AI-Powered Product Description Generator (keeping things very simple!): a tool that helps eCommerce sellers generate fully optimized product listings. Here's what it does: šŸ“ Takes a product name, key features, an uploaded image, and a spec document as inputs šŸ›ļø Generates platform-specific listings for Amazon, Etsy, Shopify, and more šŸ” Produces SEO keywords ā€” primary, secondary, and long-tail šŸ¤– Also generates some AEO and GEO content All powered by GPT-4o, built in React, and live on GitHub Pages. šŸ‘‰ Try it here: lnkd.in/gP2bntEh What I learned building this: āœ… How to design and prompt multimodal AI API calls (text + image + document in a single request) āœ… How to structure prompts to enforce strict JSON output from LLMs āœ… How to go from an idea ā†’ a working prompt ā†’ a shipped app using vibe coding āœ… Version control and deployment via GitHub, GitHub Pages and Vercel A huge thank you to Aishwarya and Arvind for building a course that actually makes you ship things. :) I am in Week-1, so just getting started. More builds coming. šŸ”Ø #AgenticAI #GenerativeAI #VibeCode #OpenAI #eCommerce #GenAcademy #MasteringAgenticAI #AIProducts #LearningInPublic

Swati Aggarwal
Week 1 of the Gen AI Building Blocks Bootcamp asked us to build something real. I built Finance-Management ā€” an AI-powered personal finance analyser. The process taught me as much as the output did. WHAT IT DOES You upload your credit card statements ā€” PDF, CSV, Excel, even a photo. The app reads them, categorises every transaction across 14 expense buckets, and gives you: ā†’ A financial health score (0ā€“100) ā†’ Spend breakdown with 8 interactive Plotly charts ā†’ A personalised budget and savings plan ā†’ A chat interface where you ask questions about your own data ā†’ A downloadable PDF report + CSV export Across multiple currencies ā€” USD, INR, GBP, EUR and 15 others. Never force-converts between them. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” šŸ”’ THE FEATURE I'M PROUDEST OF ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Before a single character reaches any AI model ā€” the app runs a two-pass PHI masking pipeline. Pass 1 runs entirely offline on your machine: ā†’ Card numbers ā†’ [CARD-XXXX] ā†’ Account numbers ā†’ [ACCOUNT-XXXX] ā†’ Routing numbers ā†’ [ROUTING-XXXX] ā†’ IFSC codes ā†’ [IFSC-XXXX] ā†’ UPI IDs ā†’ [UPI-XXXX] ā†’ IBAN, SWIFT, CVV ā€” all tokenised Pass 2 sends only the already-masked text to Groq ā€” a single batched call ā€” to catch anything regex missed. The user sees a green badge: "6 sensitive identifiers removed before sending to AI." Privacy isn't a checkbox. It's a design constraint that shapes everything else. STACK ā†’ Streamlit (4-tab UI: Upload / Dashboard / Plan / Chat) ā†’ UV for package management ā†’ Groq ā€” llama-3.3-70b-versatile for analysis, plan & chat ā†’ Groq ā€” llama-3.1-8b-instant for PHI validation & dashboard summary ā†’ Plotly + reportlab for charts and PDF export ā†’ Entirely on Groq free tier ā€” no credit card, no paid models šŸ”— lnkd.in/gaseXDXQ Repo includes full source, mock statements for testing, the vibe coding prompt I used, and setup instructions. Thank you to Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen Academy The agentic mental map they taught ā€” domain knowledge ā†’ goals ā†’ workflow ā†’ decisions ā†’ actions ā†’ autonomy ā€” I used it before writing a single line of code. The PHI architecture, model routing, lazy-load pattern ā€” all of it came from applying that framework to a real problem. That's what good teaching does. It gives you a way of thinking that travels with you. Week 2 is RAG. I'll keep sharing. #AIAgents #GenerativeAI #LLM #AgenticAI #Streamlit #Groq #LLMApps #HealthcareAI #CloudSecurity #FinTech #VibeCoding #AIBootcamp #Python #PrivacyByDesign #PersonalFinance
Week 1 of the Gen AI Building Blocks Bootcamp asked us to build something real. I built Finance-Management ā€” an AI-powered personal finance analyser. The process taught me as much as the output did. WHAT IT DOES You upload your credit card statements ā€” PDF, CSV, Excel, even a photo. The app reads them, categorises every transaction across 14 expense buckets, and gives you: ā†’ A financial health score (0ā€“100) ā†’ Spend breakdown with 8 interactive Plotly charts ā†’ A personalised budget and savings plan ā†’ A chat interface where you ask questions about your own data ā†’ A downloadable PDF report + CSV export Across multiple currencies ā€” USD, INR, GBP, EUR and 15 others. Never force-converts between them. ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” šŸ”’ THE FEATURE I'M PROUDEST OF ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā”ā” Before a single character reaches any AI model ā€” the app runs a two-pass PHI masking pipeline. Pass 1 runs entirely offline on your machine: ā†’ Card numbers ā†’ [CARD-XXXX] ā†’ Account numbers ā†’ [ACCOUNT-XXXX] ā†’ Routing numbers ā†’ [ROUTING-XXXX] ā†’ IFSC codes ā†’ [IFSC-XXXX] ā†’ UPI IDs ā†’ [UPI-XXXX] ā†’ IBAN, SWIFT, CVV ā€” all tokenised Pass 2 sends only the already-masked text to Groq ā€” a single batched call ā€” to catch anything regex missed. The user sees a green badge: "6 sensitive identifiers removed before sending to AI." Privacy isn't a checkbox. It's a design constraint that shapes everything else. STACK ā†’ Streamlit (4-tab UI: Upload / Dashboard / Plan / Chat) ā†’ UV for package management ā†’ Groq ā€” llama-3.3-70b-versatile for analysis, plan & chat ā†’ Groq ā€” llama-3.1-8b-instant for PHI validation & dashboard summary ā†’ Plotly + reportlab for charts and PDF export ā†’ Entirely on Groq free tier ā€” no credit card, no paid models šŸ”— lnkd.in/gaseXDXQ Repo includes full source, mock statements for testing, the vibe coding prompt I used, and setup instructions. Thank you to Aishwarya Srinivasan and Arvind Narayanamurthy at The Gen Academy The agentic mental map they taught ā€” domain knowledge ā†’ goals ā†’ workflow ā†’ decisions ā†’ actions ā†’ autonomy ā€” I used it before writing a single line of code. The PHI architecture, model routing, lazy-load pattern ā€” all of it came from applying that framework to a real problem. That's what good teaching does. It gives you a way of thinking that travels with you. Week 2 is RAG. I'll keep sharing. #AIAgents #GenerativeAI #LLM #AgenticAI #Streamlit #Groq #LLMApps #HealthcareAI #CloudSecurity #FinTech #VibeCoding #AIBootcamp #Python #PrivacyByDesign #PersonalFinance

Mujahid Mirza
Closing out Week 1 of Mastering Agentic AI šŸŽÆ and if I had to pick one key takeaway, it's this: "Attention Is All You Need" - the 2017 paper every LLM today is built on.šŸ“Ž lnkd.in/gnsEPtMm I broke it down using the simplest possible example: "What is the capital of France?" The image below captures the full pipeline visually and here's what's actually happening step by step šŸ‘‡ 1ļøāƒ£ Splits the sentence into tokens ā€” not words, chunks. "capital" becomes "cap" + "ital" 2ļøāƒ£ Converts each token into thousands of numbers that encode its meaning 3ļøāƒ£ Passes those numbers through stacked Transformer layers ā€” at each layer, every token asks every other token: "how relevant are you to me?" 4ļøāƒ£ "capital" and "France" lock onto each other ā€” not because they're next to each other, but because they're mathematically relevant. That's attention. šŸ” 5ļøāƒ£ The answer builds one token at a time: "The"... "capital"... "of"... "France"... "is"... "Paris" 6ļøāƒ£ Tokens are assembled back into text and streamed to your screen The model doesn't read left to right like a human. It looks at every word simultaneously and scores relevance between every pair. That's why context window size matters. That's why prompt structure matters. You can't engineer what you don't understand. Week 2 goes deeper - RAG and context engineering are up next. Given everything this week surfaced, the timing couldn't be better. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #AgenticAI #MasteringAgenticAI #Transformers #LLM #RAG #ContextEngineering
Closing out Week 1 of Mastering Agentic AI šŸŽÆ and if I had to pick one key takeaway, it's this: "Attention Is All You Need" - the 2017 paper every LLM today is built on.šŸ“Ž lnkd.in/gnsEPtMm I broke it down using the simplest possible example: "What is the capital of France?" The image below captures the full pipeline visually and here's what's actually happening step by step šŸ‘‡ 1ļøāƒ£ Splits the sentence into tokens ā€” not words, chunks. "capital" becomes "cap" + "ital" 2ļøāƒ£ Converts each token into thousands of numbers that encode its meaning 3ļøāƒ£ Passes those numbers through stacked Transformer layers ā€” at each layer, every token asks every other token: "how relevant are you to me?" 4ļøāƒ£ "capital" and "France" lock onto each other ā€” not because they're next to each other, but because they're mathematically relevant. That's attention. šŸ” 5ļøāƒ£ The answer builds one token at a time: "The"... "capital"... "of"... "France"... "is"... "Paris" 6ļøāƒ£ Tokens are assembled back into text and streamed to your screen The model doesn't read left to right like a human. It looks at every word simultaneously and scores relevance between every pair. That's why context window size matters. That's why prompt structure matters. You can't engineer what you don't understand. Week 2 goes deeper - RAG and context engineering are up next. Given everything this week surfaced, the timing couldn't be better. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #AgenticAI #MasteringAgenticAI #Transformers #LLM #RAG #ContextEngineering

Sujeet Kumar
I always wondered ā€” when I type something to an AI, how does it actually understand me? Like, does it read words the way we do? Does it have some kind of internal dictionary? Turns out... it's something far more fascinating. And honestly, once I understood it, I couldn't stop thinking about it. Here's what actually happens under the hood šŸ‘‡ šŸ­. Your words get broken into tokens Before the AI reads anything, it slices your sentence into small chunks called tokens ā€” roughly one word each. Every token gets a unique ID number. "How do planes fly?" becomes [4012, 278, 9141, 1823, 30]. Just numbers. šŸ®. Every word looks at every other word This is where it gets wild. The model doesn't read left to right like we do. Every single token scores every other token ā€” asking "how relevant are you to understanding me right now?" This is called attention, and it's why the AI knows that "fly" matters more to "planes" than "the" does. šŸÆ. This repeats across dozens of layers Each layer passes what it learned to the next ā€” building from grammar ā†’ word relationships ā†’ meaning ā†’ intent. By the final layer, the model doesn't just know what your words mean. It knows what you're trying to ask. šŸ°. It writes the answer one word at a time The model predicts the single most likely next word. Then the next. Then the next ā€” until a full, coherent response appears. No dictionary. No lookup table. Just patterns, numbers, and billions of learned weights. The GIF walks through all 4 steps visually ā€” I made it because I couldn't find a simple enough explanation. If you've ever talked to ChatGPT, Claude, or any AI assistant ā€” this is exactly what happened in the background, in milliseconds. Mind = šŸ¤Æ Aishwarya Srinivasan Arvind Narayanamurthy #ArtificialIntelligence #MachineLearning #AI #HowAIWorks #Tech #LLM
I always wondered ā€” when I type something to an AI, how does it actually understand me? Like, does it read words the way we do? Does it have some kind of internal dictionary? Turns out... it's something far more fascinating. And honestly, once I understood it, I couldn't stop thinking about it. Here's what actually happens under the hood šŸ‘‡ šŸ­. Your words get broken into tokens Before the AI reads anything, it slices your sentence into small chunks called tokens ā€” roughly one word each. Every token gets a unique ID number. "How do planes fly?" becomes [4012, 278, 9141, 1823, 30]. Just numbers. šŸ®. Every word looks at every other word This is where it gets wild. The model doesn't read left to right like we do. Every single token scores every other token ā€” asking "how relevant are you to understanding me right now?" This is called attention, and it's why the AI knows that "fly" matters more to "planes" than "the" does. šŸÆ. This repeats across dozens of layers Each layer passes what it learned to the next ā€” building from grammar ā†’ word relationships ā†’ meaning ā†’ intent. By the final layer, the model doesn't just know what your words mean. It knows what you're trying to ask. šŸ°. It writes the answer one word at a time The model predicts the single most likely next word. Then the next. Then the next ā€” until a full, coherent response appears. No dictionary. No lookup table. Just patterns, numbers, and billions of learned weights. The GIF walks through all 4 steps visually ā€” I made it because I couldn't find a simple enough explanation. If you've ever talked to ChatGPT, Claude, or any AI assistant ā€” this is exactly what happened in the background, in milliseconds. Mind = šŸ¤Æ Aishwarya Srinivasan Arvind Narayanamurthy #ArtificialIntelligence #MachineLearning #AI #HowAIWorks #Tech #LLM

Navneet Singh
Just attended an incredible session on post-training and agentic AI by Chris Alexiuk from NVIDIA, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy. My biggest takeaway? Post-training is where the magic happens. Pretraining gives a model knowledge. But post-training teaches it how to behave. That is the shift that took us from- Autocomplete machines ā†’ chat assistants ā†’ agentic models And now, systems like Nemotron-3 are pushing that even further. One idea from the session really stood out: A real agent is not just a chatbot. A real agent runs in a loop: goal + context ā†’ action ā†’ tool call ā†’ observe ā†’ update plan ā†’ repeat That loop matters just as much as the model itself. Because intelligence is not only about what the model knows. It is about what the system can reliably do. Another powerful idea: Reward can be a program. Instead of relying only on trained reward models, programmatic verifiers can check outputs using tools like: ā€¢ pytest ā€¢ SymPy ā€¢ JSONSchema That makes reward signals cheaper, more auditable, and less likely to drift. This is huge for agentic systems. The infrastructure side was just as impressive. Scaling post-training is not only a research problem. It is an engineering problem. You need: ā€¢ async rollouts ā€¢ sandboxed environments ā€¢ scalable evaluation ā€¢ reliable verifiers ā€¢ strong execution pipelines That is what makes agentic AI production-ready. The key message I took away: There is no wall in post-training. Better data, better environments, better algorithms, and better infrastructure can keep pushing model performance forward. The future of AI will not be built only by people who train bigger models. It will be built by people who understand how to shape model behavior after pretraining. Post-training is becoming the real battlefield. #LLM #ReinforcementLearning #NVIDIA #AgenticAI #PostTraining #MachineLearning #AI #DeepLearning
Just attended an incredible session on post-training and agentic AI by Chris Alexiuk from NVIDIA, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy. My biggest takeaway? Post-training is where the magic happens. Pretraining gives a model knowledge. But post-training teaches it how to behave. That is the shift that took us from- Autocomplete machines ā†’ chat assistants ā†’ agentic models And now, systems like Nemotron-3 are pushing that even further. One idea from the session really stood out: A real agent is not just a chatbot. A real agent runs in a loop: goal + context ā†’ action ā†’ tool call ā†’ observe ā†’ update plan ā†’ repeat That loop matters just as much as the model itself. Because intelligence is not only about what the model knows. It is about what the system can reliably do. Another powerful idea: Reward can be a program. Instead of relying only on trained reward models, programmatic verifiers can check outputs using tools like: ā€¢ pytest ā€¢ SymPy ā€¢ JSONSchema That makes reward signals cheaper, more auditable, and less likely to drift. This is huge for agentic systems. The infrastructure side was just as impressive. Scaling post-training is not only a research problem. It is an engineering problem. You need: ā€¢ async rollouts ā€¢ sandboxed environments ā€¢ scalable evaluation ā€¢ reliable verifiers ā€¢ strong execution pipelines That is what makes agentic AI production-ready. The key message I took away: There is no wall in post-training. Better data, better environments, better algorithms, and better infrastructure can keep pushing model performance forward. The future of AI will not be built only by people who train bigger models. It will be built by people who understand how to shape model behavior after pretraining. Post-training is becoming the real battlefield. #LLM #ReinforcementLearning #NVIDIA #AgenticAI #PostTraining #MachineLearning #AI #DeepLearning

Avishek Bhattacharjee
šŸ† Can data predict the 2026 FIFA World Cup winner? I analyzed team-level data from six previous World Cups (2002ā€“2022), alongside a separate 48-team prediction dataset for 2026. The data was sourced from Kaggle, and the complete analysis, predictive modeling, validation, and sports analytics dashboard were developed using Codex. šŸ”® The modelā€™s 2026 predictions: šŸ„‡ Champion: Argentina šŸ„ˆ Runner-up: Germany šŸ”„ Semi-finalists: Argentina, Germany, Brazil and Spain šŸŽÆ Other projected quarter-finalists: France, England, Uruguay šŸ‘€ Dark horses: Portugal, Netherlands The model considered: āš½ Recent wins, losses and goal difference šŸ“Š FIFA rankings and points šŸ’° Squad market value šŸŸļø Host advantage šŸ† World Cup experience and knockout pedigree šŸŽ‚ Average squad age I also created an interactive-style Sports Analytics Dashboard covering team performance forecasts, advancement probabilities, tournament-stage classifications, and model validation. Of course, football is gloriously unpredictable, but that is exactly what makes combining data and sport so fascinating. Who is your pick to lift the trophy in 2026? Thanks, Dominik Kundel Aishwarya Srinivasan Arvind Narayanamurthy #FIFAWorldCup #WorldCup2026 #SportsAnalytics #DataAnalytics #DataScience #MachineLearning #Codex #Kaggle #FootballAnalytics #PredictiveAnalytics #DashboardDesign
šŸ† Can data predict the 2026 FIFA World Cup winner? I analyzed team-level data from six previous World Cups (2002ā€“2022), alongside a separate 48-team prediction dataset for 2026. The data was sourced from Kaggle, and the complete analysis, predictive modeling, validation, and sports analytics dashboard were developed using Codex. šŸ”® The modelā€™s 2026 predictions: šŸ„‡ Champion: Argentina šŸ„ˆ Runner-up: Germany šŸ”„ Semi-finalists: Argentina, Germany, Brazil and Spain šŸŽÆ Other projected quarter-finalists: France, England, Uruguay šŸ‘€ Dark horses: Portugal, Netherlands The model considered: āš½ Recent wins, losses and goal difference šŸ“Š FIFA rankings and points šŸ’° Squad market value šŸŸļø Host advantage šŸ† World Cup experience and knockout pedigree šŸŽ‚ Average squad age I also created an interactive-style Sports Analytics Dashboard covering team performance forecasts, advancement probabilities, tournament-stage classifications, and model validation. Of course, football is gloriously unpredictable, but that is exactly what makes combining data and sport so fascinating. Who is your pick to lift the trophy in 2026? Thanks, Dominik Kundel Aishwarya Srinivasan Arvind Narayanamurthy #FIFAWorldCup #WorldCup2026 #SportsAnalytics #DataAnalytics #DataScience #MachineLearning #Codex #Kaggle #FootballAnalytics #PredictiveAnalytics #DashboardDesign

Utthra Kaushik
It's only Week 1 of my Agentic AI course and I already built an app I'm genuinely excited to use for myself. 2-3 hours. A working personal finance tracker. And the most important decision I made had nothing to do with code. I started the way most people do ā€” opened Claude Code and just asked it to build the app. It produced something. Technically impressive. But I had no way to steer it, and it wasn't what I actually wanted. Then the PM brain kicked in. Stop. Go back. Define the problem properly first. So I spent time with Claude just talking through the specs: ā€” What accounts do I have? ā€” What's the difference between spending money and moving it between accounts? ā€” What's V0 vs what goes on the roadmap? ā€” What am I not thinking about? That last question alone surfaced 10 things I hadn't considered. Once the spec was locked, the actual build prompt was straightforward. The app ā€” fintrack ā€” solves something that has always frustrated me: most finance apps misinterpret transfers as income or expense. fintrack automatically detects those transfers and correctly buckets the spending categories of that payment. V0 features: ā†’ Multi-account support across checking, savings, credit cards, investments ā†’ CSV import from any bank with saved format profiles ā†’ Automatic transfer matching across accounts ā†’ Custom dashboard with net worth, net in/out, time series, spending by category V1 is already specced ā€” Same flows, real persistence. Code + PRD on GitHub šŸ‘‡ lnkd.in/eKh_dRsJ Thank you to Aishwarya Srinivasan and Arvind for building this amazing courseā€” 2 lectures, 3 guest lectures, and office hours in Week 1 alone. What an enriching start. #AgenticAI #BuildInPublic #AI #ProductManagement
It's only Week 1 of my Agentic AI course and I already built an app I'm genuinely excited to use for myself. 2-3 hours. A working personal finance tracker. And the most important decision I made had nothing to do with code. I started the way most people do ā€” opened Claude Code and just asked it to build the app. It produced something. Technically impressive. But I had no way to steer it, and it wasn't what I actually wanted. Then the PM brain kicked in. Stop. Go back. Define the problem properly first. So I spent time with Claude just talking through the specs: ā€” What accounts do I have? ā€” What's the difference between spending money and moving it between accounts? ā€” What's V0 vs what goes on the roadmap? ā€” What am I not thinking about? That last question alone surfaced 10 things I hadn't considered. Once the spec was locked, the actual build prompt was straightforward. The app ā€” fintrack ā€” solves something that has always frustrated me: most finance apps misinterpret transfers as income or expense. fintrack automatically detects those transfers and correctly buckets the spending categories of that payment. V0 features: ā†’ Multi-account support across checking, savings, credit cards, investments ā†’ CSV import from any bank with saved format profiles ā†’ Automatic transfer matching across accounts ā†’ Custom dashboard with net worth, net in/out, time series, spending by category V1 is already specced ā€” Same flows, real persistence. Code + PRD on GitHub šŸ‘‡ lnkd.in/eKh_dRsJ Thank you to Aishwarya Srinivasan and Arvind for building this amazing courseā€” 2 lectures, 3 guest lectures, and office hours in Week 1 alone. What an enriching start. #AgenticAI #BuildInPublic #AI #ProductManagement

Jayapraba Mohan
A YouTube video from Aishwarya Srinivasan made me rethink a skill I've been underestimating for years. She explained Python data types using simple analogies: š‘Ø š’š’Šš’”š’• š’Šš’” š’‚ š’š’š’„š’Œš’†š’“ - š’‘š’–š’• š’•š’‰š’Šš’š’ˆš’” š’Šš’, š’•š’‚š’Œš’† š’•š’‰š’Šš’š’ˆš’” š’š’–š’•. š‘Ø š’…š’Šš’„š’•š’Šš’š’š’‚š’“š’š š’Šš’” š’‚ š’‘š’‰š’š’š’† š’ƒš’š’š’Œ -š’š’š’š’Œ š’–š’‘ š’ƒš’š š’š’‚š’Žš’†. š‘Ø š’•š’–š’‘š’š’† š’Šš’” š’‚ š’”š’†š’‚š’š’†š’… š’†š’š’—š’†š’š’š’‘š’† - š’š’š’„š’† š’„š’š’š’”š’†š’…, š’Šš’• š’…š’š’†š’”š’'š’• š’„š’‰š’‚š’š’ˆš’†. š‘Ø š’”š’†š’• š’Šš’” š’‚ š’ˆš’–š’†š’”š’• š’š’Šš’”š’• - š’š’ š’…š’–š’‘š’š’Šš’„š’‚š’•š’†š’” š’‚š’š’š’š’˜š’†š’…. I've spent 15 years working in data and engineering. I already knew what lists, dictionaries, tuples, and sets were. But that explanation stuck with me more than any technical definition ever had. And it made me realize something. We're entering a world where knowledge is becoming abundant. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜Øš˜¦š˜Æš˜¦š˜³š˜¢š˜µš˜¦ š˜¤š˜°š˜„š˜¦. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜¦š˜¹š˜±š˜­š˜¢š˜Ŗš˜Æ š˜¤š˜°š˜Æš˜¤š˜¦š˜±š˜µš˜“. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜¢š˜Æš˜“š˜øš˜¦š˜³ š˜²š˜¶š˜¦š˜“š˜µš˜Ŗš˜°š˜Æš˜“. The advantage is no longer having information. The advantage is making information understandable. For years, technical depth was enough. Now technical depth plus communication is becoming a superpower. The people who thrive in the AI era won't just be the ones who understand technology. They'll be the ones who can make it click for everyone else. šˆš§ šš š°šØš«š„š š°š”šžš«šž š€šˆ šœššš§ š šžš§šžš«ššš­šž š¢š§šŸšØš«š¦ššš­š¢šØš§, š­š”šž ššš›š¢š„š¢š­š² š­šØ š¦ššš¤šž š¢š§šŸšØš«š¦ššš­š¢šØš§ š®š§ššžš«š¬š­ššš§šššš›š„šž š¦ššš² š›šžšœšØš¦šž š­š”šž š¦šØš¬š­ šÆššš„š®ššš›š„šž š¬š¤š¢š„š„ šØšŸ ššš„š„. #DataEngineering #Python #ArtificialIntelligence #Communication
A YouTube video from Aishwarya Srinivasan made me rethink a skill I've been underestimating for years. She explained Python data types using simple analogies: š‘Ø š’š’Šš’”š’• š’Šš’” š’‚ š’š’š’„š’Œš’†š’“ - š’‘š’–š’• š’•š’‰š’Šš’š’ˆš’” š’Šš’, š’•š’‚š’Œš’† š’•š’‰š’Šš’š’ˆš’” š’š’–š’•. š‘Ø š’…š’Šš’„š’•š’Šš’š’š’‚š’“š’š š’Šš’” š’‚ š’‘š’‰š’š’š’† š’ƒš’š’š’Œ -š’š’š’š’Œ š’–š’‘ š’ƒš’š š’š’‚š’Žš’†. š‘Ø š’•š’–š’‘š’š’† š’Šš’” š’‚ š’”š’†š’‚š’š’†š’… š’†š’š’—š’†š’š’š’‘š’† - š’š’š’„š’† š’„š’š’š’”š’†š’…, š’Šš’• š’…š’š’†š’”š’'š’• š’„š’‰š’‚š’š’ˆš’†. š‘Ø š’”š’†š’• š’Šš’” š’‚ š’ˆš’–š’†š’”š’• š’š’Šš’”š’• - š’š’ š’…š’–š’‘š’š’Šš’„š’‚š’•š’†š’” š’‚š’š’š’š’˜š’†š’…. I've spent 15 years working in data and engineering. I already knew what lists, dictionaries, tuples, and sets were. But that explanation stuck with me more than any technical definition ever had. And it made me realize something. We're entering a world where knowledge is becoming abundant. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜Øš˜¦š˜Æš˜¦š˜³š˜¢š˜µš˜¦ š˜¤š˜°š˜„š˜¦. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜¦š˜¹š˜±š˜­š˜¢š˜Ŗš˜Æ š˜¤š˜°š˜Æš˜¤š˜¦š˜±š˜µš˜“. š˜ˆš˜ š˜¤š˜¢š˜Æ š˜¢š˜Æš˜“š˜øš˜¦š˜³ š˜²š˜¶š˜¦š˜“š˜µš˜Ŗš˜°š˜Æš˜“. The advantage is no longer having information. The advantage is making information understandable. For years, technical depth was enough. Now technical depth plus communication is becoming a superpower. The people who thrive in the AI era won't just be the ones who understand technology. They'll be the ones who can make it click for everyone else. šˆš§ šš š°šØš«š„š š°š”šžš«šž š€šˆ šœššš§ š šžš§šžš«ššš­šž š¢š§šŸšØš«š¦ššš­š¢šØš§, š­š”šž ššš›š¢š„š¢š­š² š­šØ š¦ššš¤šž š¢š§šŸšØš«š¦ššš­š¢šØš§ š®š§ššžš«š¬š­ššš§šššš›š„šž š¦ššš² š›šžšœšØš¦šž š­š”šž š¦šØš¬š­ šÆššš„š®ššš›š„šž š¬š¤š¢š„š„ šØšŸ ššš„š„. #DataEngineering #Python #ArtificialIntelligence #Communication

khyathi koteru
Attended a great guest lecture by Dominik Kundel from OpenAI on how coding agents like Codex are evolving beyond autocomplete and pair programming into fully delegated agents. The biggest takeaway: getting value from agents is not just about writing perfect prompts. It is about giving them the right context, validation tools, and verification workflows. Dom shared how agents can use connected context from tools like Slack, Google Drive, Gmail, Calendar, GitHub, and docs to work more like a teammate than a simple assistant. He also emphasized the importance of linters, tests, previews, review artifacts, and feedback loops so agents can validate their own work and make human review easier. Really insightful session on how agentic workflows can support not just engineering, but also content, documentation, marketing, operations, and everyday productivity. Thank you, Dominik Kundel, for sharing practical examples and best practices. A huge thank you to Aishwarya Srinivasan and Arvind Narayanamurthy for organizing and hosting an outstanding guest lecture as part of the Mastering Agentic AI program.
Attended a great guest lecture by Dominik Kundel from OpenAI on how coding agents like Codex are evolving beyond autocomplete and pair programming into fully delegated agents. The biggest takeaway: getting value from agents is not just about writing perfect prompts. It is about giving them the right context, validation tools, and verification workflows. Dom shared how agents can use connected context from tools like Slack, Google Drive, Gmail, Calendar, GitHub, and docs to work more like a teammate than a simple assistant. He also emphasized the importance of linters, tests, previews, review artifacts, and feedback loops so agents can validate their own work and make human review easier. Really insightful session on how agentic workflows can support not just engineering, but also content, documentation, marketing, operations, and everyday productivity. Thank you, Dominik Kundel, for sharing practical examples and best practices. A huge thank you to Aishwarya Srinivasan and Arvind Narayanamurthy for organizing and hosting an outstanding guest lecture as part of the Mastering Agentic AI program.

Aishwarya Kote
Built my first AI-powered Streamlit application! šŸ¤© As part of Week 1 of the Mastering Agentic AI Maven course, I built an ESG Greenwashing Detector ā€“ a Streamlit application that analyses ESG disclosure data and helps identify potential greenwashing risks across companies and sectors. When I was thinking about what application to build, I wanted to connect it to something I am currently studying. Since one of my modules is Responsible International Business and Strategic Risk Management, I decided to explore the topic of ESG reporting and greenwashing. This project gave me an opportunity to combine concepts from my studies with AI-powered application development. The application includes: -ESG data upload and analysis -Greenwashing risk dashboard -Company ESG deep-dive analysis -AI-generated ESG assessments -AI-powered ESG analyst chat using Groq Coming from a non-coding background, I honestly didn't expect to build and deploy an application within the first week. This experience showed me how powerful AI-assisted development can be and how quickly ideas can be turned into functional applications. A big thank you to Aishwarya Srinivasan, Arvind Narayanamurthy and the team for the excellent training sessions, detailed documentation, and support. ā¤ļø This is just our Week 1 learning project, and I'm excited to continue building more in the coming weeks. šŸ˜Š Live App: lnkd.in/ew_-Cr5T GitHub Repository: lnkd.in/ewvhVWeT
Built my first AI-powered Streamlit application! šŸ¤© As part of Week 1 of the Mastering Agentic AI Maven course, I built an ESG Greenwashing Detector ā€“ a Streamlit application that analyses ESG disclosure data and helps identify potential greenwashing risks across companies and sectors. When I was thinking about what application to build, I wanted to connect it to something I am currently studying. Since one of my modules is Responsible International Business and Strategic Risk Management, I decided to explore the topic of ESG reporting and greenwashing. This project gave me an opportunity to combine concepts from my studies with AI-powered application development. The application includes: -ESG data upload and analysis -Greenwashing risk dashboard -Company ESG deep-dive analysis -AI-generated ESG assessments -AI-powered ESG analyst chat using Groq Coming from a non-coding background, I honestly didn't expect to build and deploy an application within the first week. This experience showed me how powerful AI-assisted development can be and how quickly ideas can be turned into functional applications. A big thank you to Aishwarya Srinivasan, Arvind Narayanamurthy and the team for the excellent training sessions, detailed documentation, and support. ā¤ļø This is just our Week 1 learning project, and I'm excited to continue building more in the coming weeks. šŸ˜Š Live App: lnkd.in/ew_-Cr5T GitHub Repository: lnkd.in/ewvhVWeT

Ravi Uppalapati
A little while ago I shared the first version of a project from the Mastering AI Agents course ā€” an interactive world map of what every country grows. Since then I've kept iterating ā€” and that continuous building is exactly where the real learning is happening. Every new version teaches me something I couldn't have picked up any other way. What's new this time: ā€¢ A cleaner, tabbed UI with summary cards and a richer, easier-to-read map ā€¢ A built-in AI assistant that answers questions about the data ā€” "Which countries produced the most wheat in 1985?" ā€¢ Built with both Claude and Codex ā€” they complement each other wonderfully ā€¢ The Superpowers plugin made Codex even more capable for multi-step building ā€¢ Designed to stay free and efficient: a focused data context, tool-based queries, and a smart provider fallback (Gemini ā†’ Groq ā†’ Cerebras ā†’ OpenRouter ā†’ Mistral) See how it's evolved šŸ‘‡ šŸ”¹ First version: lnkd.in/euUWKc-r šŸ”ø Latest version: lnkd.in/erqBKhH7 šŸ’» Code: lnkd.in/eAk46ifj Truly grateful to Aishwarya Srinivasan, Arvind Narayanamurthy, and The Gen Academy ā€” the encouragement to keep building and sharing in public makes all the difference. Excited for what's next. šŸŒ± #AIAgents #VibeCoding #BuildInPublic #MasteringAIAgents
A little while ago I shared the first version of a project from the Mastering AI Agents course ā€” an interactive world map of what every country grows. Since then I've kept iterating ā€” and that continuous building is exactly where the real learning is happening. Every new version teaches me something I couldn't have picked up any other way. What's new this time: ā€¢ A cleaner, tabbed UI with summary cards and a richer, easier-to-read map ā€¢ A built-in AI assistant that answers questions about the data ā€” "Which countries produced the most wheat in 1985?" ā€¢ Built with both Claude and Codex ā€” they complement each other wonderfully ā€¢ The Superpowers plugin made Codex even more capable for multi-step building ā€¢ Designed to stay free and efficient: a focused data context, tool-based queries, and a smart provider fallback (Gemini ā†’ Groq ā†’ Cerebras ā†’ OpenRouter ā†’ Mistral) See how it's evolved šŸ‘‡ šŸ”¹ First version: lnkd.in/euUWKc-r šŸ”ø Latest version: lnkd.in/erqBKhH7 šŸ’» Code: lnkd.in/eAk46ifj Truly grateful to Aishwarya Srinivasan, Arvind Narayanamurthy, and The Gen Academy ā€” the encouragement to keep building and sharing in public makes all the difference. Excited for what's next. šŸŒ± #AIAgents #VibeCoding #BuildInPublic #MasteringAIAgents

Mujahid Mirza
The Mastering Agentic AI bootcamp at The Gen Academy has one theme: #BuildInPublic . So here's Week 1. With my background in data and analytics I wanted to build something end-to-end, production-grade, and close to that world. Meet ToyWorld šŸ§ø - a full-stack Retail Management System covering everything from point-of-sale to sales analytics to IT administration, with role-based access control throughout. Built around an imaginary toy store, but designed to plug into any small retail business from day one. The architecture diagram below gives a quick sense of how it all fits together. Backed by a star schema in Supabase with 10,000+ sales rows, 500 customers, 300+ SKUs, 2 years of data with seasonality modeling, and 16 interactive Plotly charts across four analytic dimensions. Built to feel like production. šŸ› ļø Stack: Claude Code (+ Superpowers) Ā· Streamlit Ā· Pandas Ā· Plotly Ā· Supabase Ā· Faker I'm excited about what AI-assisted development makes possible. Sharing the work as it happens feels like the most honest way to learn and if it sparks an idea or helps someone on a similar path, even better. šŸ”— GitHub: lnkd.in/ggazramp šŸš€ Live App: lnkd.in/gjXmvQjP More weeks to come. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #BuildInPublic #AgenticAI #AIAssistedCoding #ClaudeCode #Superpowers #Streamlit #Supabase #DataAndAnalytics #RetailTech #AIBootcamp #GenerativeAI #SolutionsArchitect
The Mastering Agentic AI bootcamp at The Gen Academy has one theme: #BuildInPublic . So here's Week 1. With my background in data and analytics I wanted to build something end-to-end, production-grade, and close to that world. Meet ToyWorld šŸ§ø - a full-stack Retail Management System covering everything from point-of-sale to sales analytics to IT administration, with role-based access control throughout. Built around an imaginary toy store, but designed to plug into any small retail business from day one. The architecture diagram below gives a quick sense of how it all fits together. Backed by a star schema in Supabase with 10,000+ sales rows, 500 customers, 300+ SKUs, 2 years of data with seasonality modeling, and 16 interactive Plotly charts across four analytic dimensions. Built to feel like production. šŸ› ļø Stack: Claude Code (+ Superpowers) Ā· Streamlit Ā· Pandas Ā· Plotly Ā· Supabase Ā· Faker I'm excited about what AI-assisted development makes possible. Sharing the work as it happens feels like the most honest way to learn and if it sparks an idea or helps someone on a similar path, even better. šŸ”— GitHub: lnkd.in/ggazramp šŸš€ Live App: lnkd.in/gjXmvQjP More weeks to come. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #BuildInPublic #AgenticAI #AIAssistedCoding #ClaudeCode #Superpowers #Streamlit #Supabase #DataAndAnalytics #RetailTech #AIBootcamp #GenerativeAI #SolutionsArchitect

Rehana Sheikh
A few weeks ago, I shared my journey back into the Data + AI space with a structured plan. Hereā€™s what has unfolded since then. I didnā€™t just study; I built lnkd.in/eNTK9Bp3. Watch Demo lnkd.in/e3whCMFz My passion for structured learning led me to enroll in the Mastering Agentic AI course by Gen Academy, which became a pivotal moment for me. The course not only deepened my understanding of agentic AI but also instilled in me the confidence to start vibe coding. This approach allowed me to build with AI rather than learn about it. Vibe coding involves describing what you want to an AI coding agent and iterating through conversation instead of writing every line yourself. My tool of choice has been Claude Code. My first real project was an Ayurveda Prakriti Assessment app, a web app designed to help users discover their dosha constitution and provide personalized daily routines, diet, exercise, and seasonal lifestyle guidance based on 5,000-year-old Ayurvedic wisdom. I transformed an idea into a deployed app, all on my own, with a clear vision and an AI pair programmer. This experience taught me that the engineers who will excel in the next wave are those who can think clearly, break down problems, and guide AI to execute. Vibe coding is not about writing less code; itā€™s about thinking more clearly about what youā€™re building and why. Hereā€™s my current learning stack: ā†’ Vibe coding with Claude Codeā€”building real projects, not tutorials ā†’ Mastering Agentic AI (Gen Academy)ā€”understanding agent architectures, tool use, and reasoning patterns ā†’ Strengthening Python, SQL, and data pipeline fundamentals daily ā†’ Connecting the dots between data engineering and AI-first development The career break provided clarity, and my comeback is generating momentum. If youā€™re also upskilling in AI, returning after a break, or curious about vibe coding, letā€™s connect. Iā€™m learning in public and sharing my journey along the And if you want to try the app I built: lnkd.in/eNTK9Bp3 github link : lnkd.in/e4AiTp7V Demo Video: lnkd.in/e3whCMFz More to come. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #VibeCoding #AgenticAI #ClaudeCode #DataEngineering #AI #CareerRestart #BuildInPublic #WomenInTech #LearnInPublic #Python #GenAcademy
A few weeks ago, I shared my journey back into the Data + AI space with a structured plan. Hereā€™s what has unfolded since then. I didnā€™t just study; I built lnkd.in/eNTK9Bp3. Watch Demo lnkd.in/e3whCMFz My passion for structured learning led me to enroll in the Mastering Agentic AI course by Gen Academy, which became a pivotal moment for me. The course not only deepened my understanding of agentic AI but also instilled in me the confidence to start vibe coding. This approach allowed me to build with AI rather than learn about it. Vibe coding involves describing what you want to an AI coding agent and iterating through conversation instead of writing every line yourself. My tool of choice has been Claude Code. My first real project was an Ayurveda Prakriti Assessment app, a web app designed to help users discover their dosha constitution and provide personalized daily routines, diet, exercise, and seasonal lifestyle guidance based on 5,000-year-old Ayurvedic wisdom. I transformed an idea into a deployed app, all on my own, with a clear vision and an AI pair programmer. This experience taught me that the engineers who will excel in the next wave are those who can think clearly, break down problems, and guide AI to execute. Vibe coding is not about writing less code; itā€™s about thinking more clearly about what youā€™re building and why. Hereā€™s my current learning stack: ā†’ Vibe coding with Claude Codeā€”building real projects, not tutorials ā†’ Mastering Agentic AI (Gen Academy)ā€”understanding agent architectures, tool use, and reasoning patterns ā†’ Strengthening Python, SQL, and data pipeline fundamentals daily ā†’ Connecting the dots between data engineering and AI-first development The career break provided clarity, and my comeback is generating momentum. If youā€™re also upskilling in AI, returning after a break, or curious about vibe coding, letā€™s connect. Iā€™m learning in public and sharing my journey along the And if you want to try the app I built: lnkd.in/eNTK9Bp3 github link : lnkd.in/e4AiTp7V Demo Video: lnkd.in/e3whCMFz More to come. šŸš€ Aishwarya Srinivasan Arvind Narayanamurthy #VibeCoding #AgenticAI #ClaudeCode #DataEngineering #AI #CareerRestart #BuildInPublic #WomenInTech #LearnInPublic #Python #GenAcademy

Avishek Bhattacharjee
Small win, nothing fancy šŸŽ‰ I deployed StartupScope AI on Hugging Face Spaces ā€” an amateur startup idea validation tool built with LangChain, Streamlit, and OpenAI. This was a simple project to learn how Hugging Face works and how quickly I can go from an idea to a deployed application. What I have used: LangChain chains and prompts Streamlit for the frontend OpenAI for idea analysis Hugging Face Spaces for deployment There's always something satisfying about taking a project from a local machine to something that anyone can access and use. Aishwarya Srinivasan Arvind Narayanamurthy #AIEngineering #LangChain #Streamlit #OpenAI #HuggingFace
Small win, nothing fancy šŸŽ‰ I deployed StartupScope AI on Hugging Face Spaces ā€” an amateur startup idea validation tool built with LangChain, Streamlit, and OpenAI. This was a simple project to learn how Hugging Face works and how quickly I can go from an idea to a deployed application. What I have used: LangChain chains and prompts Streamlit for the frontend OpenAI for idea analysis Hugging Face Spaces for deployment There's always something satisfying about taking a project from a local machine to something that anyone can access and use. Aishwarya Srinivasan Arvind Narayanamurthy #AIEngineering #LangChain #Streamlit #OpenAI #HuggingFace

Navneet Singh
Two years ago, I moved to the United States and bought my first car. A few months in, something felt off. I took it to a mechanic. They plugged in a scanner, spent 10 minutes looking at it, and handed me a $200 diagnosis bill. Then they told me I needed a new catalytic converter, a $2,000 repair. I didnā€™t know enough to push back. I didnā€™t know what the data meant. I just had to trust them. That moment stuck with me. So for my Week 1 project at the Mastering Agentic AI Bootcamp, I didnā€™t build the standard stock portfolio analyzer. I built the tool I wish I had that day. The š—¢š—•š——šŸ® š—˜š—»š—“š—¶š—»š—² š——š—¶š—®š—“š—»š—¼š˜€š˜š—¶š—°š˜€ š—§š—²š—¹š—²š—ŗš—²š˜š—暝˜† š—Ÿš—®š—Æ - a dashboard that reads the same sensor data a professional scan tool reads and explains, in plain English, what is wrong with your car and what you actually need to fix. š—Ŗš—µš—®š˜ š—¶š˜ š—±š—¼š—²š˜€: ā†’ Detects vacuum leaks, rich/lean conditions, and fuel trim anomalies ā†’ Shows where the fault appears - idle, acceleration, or cruise ā†’ Assesses catalytic converter health from Oā‚‚ sensor waveforms ā†’ Cross-references faults with NHTSA recall and complaint data If I had had this tool two years ago, I would have known it was a vacuum leak a $30 fix, not a catalytic converter replacement. š—Ŗš—µš—®š˜ š—œ š—¹š—²š—®š—暝—»š—²š—± š—Æš˜‚š—¶š—¹š—±š—¶š—»š—“ š—¶š˜: āœ¦ Define the problem before touching the keyboard āœ¦ Modular code survives feature additions āœ¦ src/ modules should never import Streamlit āœ¦ git checkout develop -- file.py is underrated āœ¦ If pandas downloads a .tar.gz on Streamlit Cloud, stop the build 50 commits. 3 versioned releases. 18 edge cases validated. š—•š˜‚š˜ š˜š—µš—¶š˜€ š—¶š˜€ š—·š˜‚š˜€š˜ š˜š—µš—² š˜€š˜š—®š—暝˜. A $30 OBD2 dongle can extract dozens of live sensor readings from your car. Right now, this dashboard covers fuel trim, Oā‚‚ sensors, and RPM dynamics. But modern cars have hundreds of sensors - transmission temperature, ABS, throttle position, battery health, emissions systems. The vision -> a tool that any normal person can plug into their car, upload the data, and understand what is wrong before walking into a mechanicā€™s shop. No more $200 diagnostic fees. No more replacing a $2,000 catalytic converter when the real problem is a $30 vacuum hose. This tool is free. Open source. Massive thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for building a AI Cohort that pushed me to build things that actually matter. Live app link and GitHub repo in the comments below!
Two years ago, I moved to the United States and bought my first car. A few months in, something felt off. I took it to a mechanic. They plugged in a scanner, spent 10 minutes looking at it, and handed me a $200 diagnosis bill. Then they told me I needed a new catalytic converter, a $2,000 repair. I didnā€™t know enough to push back. I didnā€™t know what the data meant. I just had to trust them. That moment stuck with me. So for my Week 1 project at the Mastering Agentic AI Bootcamp, I didnā€™t build the standard stock portfolio analyzer. I built the tool I wish I had that day. The š—¢š—•š——šŸ® š—˜š—»š—“š—¶š—»š—² š——š—¶š—®š—“š—»š—¼š˜€š˜š—¶š—°š˜€ š—§š—²š—¹š—²š—ŗš—²š˜š—暝˜† š—Ÿš—®š—Æ - a dashboard that reads the same sensor data a professional scan tool reads and explains, in plain English, what is wrong with your car and what you actually need to fix. š—Ŗš—µš—®š˜ š—¶š˜ š—±š—¼š—²š˜€: ā†’ Detects vacuum leaks, rich/lean conditions, and fuel trim anomalies ā†’ Shows where the fault appears - idle, acceleration, or cruise ā†’ Assesses catalytic converter health from Oā‚‚ sensor waveforms ā†’ Cross-references faults with NHTSA recall and complaint data If I had had this tool two years ago, I would have known it was a vacuum leak a $30 fix, not a catalytic converter replacement. š—Ŗš—µš—®š˜ š—œ š—¹š—²š—®š—暝—»š—²š—± š—Æš˜‚š—¶š—¹š—±š—¶š—»š—“ š—¶š˜: āœ¦ Define the problem before touching the keyboard āœ¦ Modular code survives feature additions āœ¦ src/ modules should never import Streamlit āœ¦ git checkout develop -- file.py is underrated āœ¦ If pandas downloads a .tar.gz on Streamlit Cloud, stop the build 50 commits. 3 versioned releases. 18 edge cases validated. š—•š˜‚š˜ š˜š—µš—¶š˜€ š—¶š˜€ š—·š˜‚š˜€š˜ š˜š—µš—² š˜€š˜š—®š—暝˜. A $30 OBD2 dongle can extract dozens of live sensor readings from your car. Right now, this dashboard covers fuel trim, Oā‚‚ sensors, and RPM dynamics. But modern cars have hundreds of sensors - transmission temperature, ABS, throttle position, battery health, emissions systems. The vision -> a tool that any normal person can plug into their car, upload the data, and understand what is wrong before walking into a mechanicā€™s shop. No more $200 diagnostic fees. No more replacing a $2,000 catalytic converter when the real problem is a $30 vacuum hose. This tool is free. Open source. Massive thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for building a AI Cohort that pushed me to build things that actually matter. Live app link and GitHub repo in the comments below!

Anushree Sharma
Attended an insightful session with Harsh Vardhan Jain on ā€œMaking AI Agents Secureā€ by Aishwarya Srinivasan along with Arvind Narayanamurthy and Harsh Nath Jha and it genuinely changed the way I think about AI security in the agentic era. As AI agents become capable of writing code, calling APIs, accessing enterprise systems, and making autonomous decisions, traditional security models are no longer enough. The discussion around agent identity abuse, non-human identity sprawl, excessive permissions, long-lived tokens, and autonomous access to infrastructure was particularly eye-opening. One slide that stood out to me highlighted how a single compromised AI agent identity could potentially cascade into access across: GitHub repositories, Slack workspaces, AWS resources, etc. The session strongly emphasized why future AI systems need: Ephemeral credentials, Just-in-time access provisioning, Fine-grained permissions, Strong governance and observability layers for agents. As someone currently working in the Application Security space while also being deeply interested in AI systems, it was exciting to see the intersection of GenAI + Security being discussed with such practical depth. The future of cybersecurity will definitely involve securing not just humans, but autonomous AI identities as well. #CyberSecurity #AISecurity #GenAI #LLM #AgenticAI
Attended an insightful session with Harsh Vardhan Jain on ā€œMaking AI Agents Secureā€ by Aishwarya Srinivasan along with Arvind Narayanamurthy and Harsh Nath Jha and it genuinely changed the way I think about AI security in the agentic era. As AI agents become capable of writing code, calling APIs, accessing enterprise systems, and making autonomous decisions, traditional security models are no longer enough. The discussion around agent identity abuse, non-human identity sprawl, excessive permissions, long-lived tokens, and autonomous access to infrastructure was particularly eye-opening. One slide that stood out to me highlighted how a single compromised AI agent identity could potentially cascade into access across: GitHub repositories, Slack workspaces, AWS resources, etc. The session strongly emphasized why future AI systems need: Ephemeral credentials, Just-in-time access provisioning, Fine-grained permissions, Strong governance and observability layers for agents. As someone currently working in the Application Security space while also being deeply interested in AI systems, it was exciting to see the intersection of GenAI + Security being discussed with such practical depth. The future of cybersecurity will definitely involve securing not just humans, but autonomous AI identities as well. #CyberSecurity #AISecurity #GenAI #LLM #AgenticAI

Sham Sundar Hassan Chikkegowda
Just wrapped up an incredibly intense session diving deep into Agentic Security and the operational reality of the Agentic SOC. The material provided a necessary framework for how we must evolve our defenses. lnkd.in/ge8k_jUU This experience perfectly illustrates the critical inflection point we are all facing: šŸ•°ļø The Past: Traditional security stacks were built for static environments, leaving massive gaps when dealing with autonomous AI agents that execute code and act across infrastructure at machine speed. šŸ§­ The Present: We see that legacy controls simply fail in the Agentic era. The real work now is moving beyond reactive alerts to architecting a true Agentic Security Meshā€”a self-healing framework for governing autonomous systems. šŸ”® The Future: The future doesn't start with just listening; it starts with building the guardrails. This realization confirms why our focus on agent-driven governance and proactive control is so essential right now. The practical takeaway? We must move from theory to applicationā€”like taking a specific SOP and automating its security controls. This is exactly the hands-on engineering required for true resilience in this new landscape. Huge appreciation to Harsh Nath Jha and the organizers Aishwarya Srinivasan , Arvind Narayanamurthy for framing this necessary evolution. The work we are doing to build these future defenses is more vital than ever. #AgenticAI #AISecurity #DevSecOps #SecurityAutomation #FutureOfDefense
Just wrapped up an incredibly intense session diving deep into Agentic Security and the operational reality of the Agentic SOC. The material provided a necessary framework for how we must evolve our defenses. lnkd.in/ge8k_jUU This experience perfectly illustrates the critical inflection point we are all facing: šŸ•°ļø The Past: Traditional security stacks were built for static environments, leaving massive gaps when dealing with autonomous AI agents that execute code and act across infrastructure at machine speed. šŸ§­ The Present: We see that legacy controls simply fail in the Agentic era. The real work now is moving beyond reactive alerts to architecting a true Agentic Security Meshā€”a self-healing framework for governing autonomous systems. šŸ”® The Future: The future doesn't start with just listening; it starts with building the guardrails. This realization confirms why our focus on agent-driven governance and proactive control is so essential right now. The practical takeaway? We must move from theory to applicationā€”like taking a specific SOP and automating its security controls. This is exactly the hands-on engineering required for true resilience in this new landscape. Huge appreciation to Harsh Nath Jha and the organizers Aishwarya Srinivasan , Arvind Narayanamurthy for framing this necessary evolution. The work we are doing to build these future defenses is more vital than ever. #AgenticAI #AISecurity #DevSecOps #SecurityAutomation #FutureOfDefense

Anand Sridharan
From Text to SQL: How SQL AI Agents are Transforming Data Access Wrapped up an incredible session on š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ with Rami Krispin, where he shared his rich experience and production ready play-book. Encapsulated - Top 5 pointers (provided AI generated image for illustration only) // Context is King, Not the Model You don't need the fanciest LLM to build powerful SQL agents. A thoughtfully engineered prompt with rich context (schema details, categorical values, business logic) can outperform expensive models. It's about š˜©š˜°š˜ø you ask, not just š˜øš˜©š˜° you ask. // The Agent Loop is Everything It's not just LLM ā†’ Output. Real agents iterate through reasoning ā†’ tool calls ā†’ observation ā†’ refinement. This dynamic execution path is what separates true agents from simple API wrappers. // Safety Guardrails Aren't Optional : They're Essential Read-only access, query validation, row limits, logs and error handling aren't afterthoughts. They're foundational. One misconfigured query can bring down your entire database. Period. // Memory Management Changes the Game Adding conversation history transforms single-shot queries into meaningful dialogue. Users can ask follow-ups naturally. But manage it wisely, unbounded memory explodes your token costs. // Evaluation & Fallback Strategies = Smart Cost Optimization Test your models systematically. Sometimes GPT-4 mini outperforms GPT-4 Turbo at 1/4 the cost. Build intelligent fallback mechanisms; simple models for routine queries, powerful ones for edge cases. A heartfelt thank you to Aishwarya Srinivasan, Arvind Narayanamurthy for setting up context/de-briefing about AI agent, and bringing in Rami Krispin (SME) for this Lightening session. Your generosity in sharing examples/use cases, live demos are highly appreciated. I gained additional insights and I'm sure all the participants who attended "live" would have benefitted, and reach out to discord community for Q&A/follow-up. #SQLAIAgents #AIEngineering #LLMs #DataAccess #ProductionAI #TechEducation #LinkedInLearning
From Text to SQL: How SQL AI Agents are Transforming Data Access Wrapped up an incredible session on š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ with Rami Krispin, where he shared his rich experience and production ready play-book. Encapsulated - Top 5 pointers (provided AI generated image for illustration only) // Context is King, Not the Model You don't need the fanciest LLM to build powerful SQL agents. A thoughtfully engineered prompt with rich context (schema details, categorical values, business logic) can outperform expensive models. It's about š˜©š˜°š˜ø you ask, not just š˜øš˜©š˜° you ask. // The Agent Loop is Everything It's not just LLM ā†’ Output. Real agents iterate through reasoning ā†’ tool calls ā†’ observation ā†’ refinement. This dynamic execution path is what separates true agents from simple API wrappers. // Safety Guardrails Aren't Optional : They're Essential Read-only access, query validation, row limits, logs and error handling aren't afterthoughts. They're foundational. One misconfigured query can bring down your entire database. Period. // Memory Management Changes the Game Adding conversation history transforms single-shot queries into meaningful dialogue. Users can ask follow-ups naturally. But manage it wisely, unbounded memory explodes your token costs. // Evaluation & Fallback Strategies = Smart Cost Optimization Test your models systematically. Sometimes GPT-4 mini outperforms GPT-4 Turbo at 1/4 the cost. Build intelligent fallback mechanisms; simple models for routine queries, powerful ones for edge cases. A heartfelt thank you to Aishwarya Srinivasan, Arvind Narayanamurthy for setting up context/de-briefing about AI agent, and bringing in Rami Krispin (SME) for this Lightening session. Your generosity in sharing examples/use cases, live demos are highly appreciated. I gained additional insights and I'm sure all the participants who attended "live" would have benefitted, and reach out to discord community for Q&A/follow-up. #SQLAIAgents #AIEngineering #LLMs #DataAccess #ProductionAI #TechEducation #LinkedInLearning

Haadvitha Nimmagadda
Attended a session on Building SQL AI Agents at The Gen Academy today. Honest take? Parts of it felt overwhelming. As someone just starting to explore AI, I could follow the high-level ideas, but the deeper discussions around context, architecture, and guardrails made one thing clear: Thereā€™s a lot to learn here. And that was the most useful part. Before writing a single line of code, the session broke down how agents actually work: ā€¢ Agents donā€™t respond in one step, they run a loop: Think ā†’ Act ā†’ Observe until the goal is met ā€¢ There are key levers you can control to shape that loop ā€¢ Every agent is built from a set of core building blocks ā€¢ Not every problem needs an agent, knowing when to use one matters Understanding these fundamentals before building matters more than I expected. Then it got more practical, what it takes to move from idea to something usable: ā€¢ The gap between a polished demo and a production system is significant ā€¢ Context is more than a prompt, it includes instructions, knowledge, and state ā€¢ Guardrails and validation are what make these systems reliable Instead of trying to learn everything upfront, Iā€™ve decided to start building a simple SQL agent and learn through the process. Thanks to Aishwarya Srinivasan, Arvind Narayanamurthy and Rami Krispin for keeping it practical and grounded. #AIAgents #GenAI #SQLAgent #DataEngineering #DataAnalytics
Attended a session on Building SQL AI Agents at The Gen Academy today. Honest take? Parts of it felt overwhelming. As someone just starting to explore AI, I could follow the high-level ideas, but the deeper discussions around context, architecture, and guardrails made one thing clear: Thereā€™s a lot to learn here. And that was the most useful part. Before writing a single line of code, the session broke down how agents actually work: ā€¢ Agents donā€™t respond in one step, they run a loop: Think ā†’ Act ā†’ Observe until the goal is met ā€¢ There are key levers you can control to shape that loop ā€¢ Every agent is built from a set of core building blocks ā€¢ Not every problem needs an agent, knowing when to use one matters Understanding these fundamentals before building matters more than I expected. Then it got more practical, what it takes to move from idea to something usable: ā€¢ The gap between a polished demo and a production system is significant ā€¢ Context is more than a prompt, it includes instructions, knowledge, and state ā€¢ Guardrails and validation are what make these systems reliable Instead of trying to learn everything upfront, Iā€™ve decided to start building a simple SQL agent and learn through the process. Thanks to Aishwarya Srinivasan, Arvind Narayanamurthy and Rami Krispin for keeping it practical and grounded. #AIAgents #GenAI #SQLAgent #DataEngineering #DataAnalytics

Anand Ganesan
Fun and learning sounds like extremities until there is passion in it. Aishwarya Srinivasan you made the evening a fun AI enlightening meetup and the truth is we genuinely learn forever. Looking forward to Cohort 2
Fun and learning sounds like extremities until there is passion in it. Aishwarya Srinivasan you made the evening a fun AI enlightening meetup and the truth is we genuinely learn forever. Looking forward to Cohort 2

Yuvansh Bhardwaj
Attended an incredible session on š—•š˜‚š—¶š—¹š—±š—¶š—»š—“ š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ hosted by Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin. š—œš—³ š—®š—“š—²š—»š˜š˜€ š—®š—暝—² š˜š—µš—² š—Æš—暝—®š—¶š—» š—¼š—³ š˜š—µš—² š—”š—œ š—²š—°š—¼š˜€š˜†š˜€š˜š—²š—ŗ, š˜š—µš—²š—» š—±š—®š˜š—® š—®š—»š—± š—®š—»š—®š—¹š˜†š˜š—¶š—°š˜€ š—®š—暝—² š˜š—µš—² š—³š˜‚š—²š—¹. So I went ahead and built my own SQL AI Agent from scratch. What I built: āœ… Natural language to SQL using Ollama (free, no API key) āœ… SQLite database with zero setup āœ… Streamlit UI with query history and CSV export āœ… Runs 100% locally, no cloud, no cost Huge thanks to Aishwarya, Arvind, and Rami for making this so hands-on. Sessions like this push you to stop watching and start building. Check out the repo here: šŸ”— lnkd.in/ebX4Dqix #AI #SQLAgent #Ollama #LLM #DataScience #ArtificialIntelligence #DataEngineering #MachineLearning #Python #Canada
Attended an incredible session on š—•š˜‚š—¶š—¹š—±š—¶š—»š—“ š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ hosted by Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin. š—œš—³ š—®š—“š—²š—»š˜š˜€ š—®š—暝—² š˜š—µš—² š—Æš—暝—®š—¶š—» š—¼š—³ š˜š—µš—² š—”š—œ š—²š—°š—¼š˜€š˜†š˜€š˜š—²š—ŗ, š˜š—µš—²š—» š—±š—®š˜š—® š—®š—»š—± š—®š—»š—®š—¹š˜†š˜š—¶š—°š˜€ š—®š—暝—² š˜š—µš—² š—³š˜‚š—²š—¹. So I went ahead and built my own SQL AI Agent from scratch. What I built: āœ… Natural language to SQL using Ollama (free, no API key) āœ… SQLite database with zero setup āœ… Streamlit UI with query history and CSV export āœ… Runs 100% locally, no cloud, no cost Huge thanks to Aishwarya, Arvind, and Rami for making this so hands-on. Sessions like this push you to stop watching and start building. Check out the repo here: šŸ”— lnkd.in/ebX4Dqix #AI #SQLAgent #Ollama #LLM #DataScience #ArtificialIntelligence #DataEngineering #MachineLearning #Python #Canada

Subhiksha Mukuntharaj
Had a great time at the AI meetup hosted by The Gen Academy in SF this week. One thing from Aishwarya Srinivasanā€™s talk that really stayed with me was how much GTM for developer tools has changed. Feels like polished marketing and generic funnels matter way less now. People just want to see the product actually work. The teams standing out are the ones shipping fast, building in public, and earning trust from developers over time. Really enjoyed the conversations and meeting more people building in the space. #AI #DataScience #GenAI #AIAgents
Had a great time at the AI meetup hosted by The Gen Academy in SF this week. One thing from Aishwarya Srinivasanā€™s talk that really stayed with me was how much GTM for developer tools has changed. Feels like polished marketing and generic funnels matter way less now. People just want to see the product actually work. The teams standing out are the ones shipping fast, building in public, and earning trust from developers over time. Really enjoyed the conversations and meeting more people building in the space. #AI #DataScience #GenAI #AIAgents

Rajiv Vittala
Great People.. Better Ideas...āš”ļø Just wrapped up an incredible day at the AI Meetup in Manhattan New York. Itā€™s a room full of builders, founders, and enthusiasts who are actually making it happen. My 3 big takeaways from the night: Speed is the new currency: The gap between "idea" and "deployed product" is shrinking by the day. Community > Code: The most powerful breakthroughs happen when we step away from the screen and start talking to each other. NYC is the place to be: The density of talent here is unmatched. Huge thanks to the organizers Aishwarya Srinivasan Arvind Narayanamurthy Aditya Suresh and everyone I chatted with today. If we met, letā€™s keep the conversation going! šŸ‘‡ Special thanks to Anna Mai from Pinecone for the incredible insights. #AI #NYCTech #ArtificialIntelligence #Networking #Innovation #GenerativeAI #Newyork
Great People.. Better Ideas...āš”ļø Just wrapped up an incredible day at the AI Meetup in Manhattan New York. Itā€™s a room full of builders, founders, and enthusiasts who are actually making it happen. My 3 big takeaways from the night: Speed is the new currency: The gap between "idea" and "deployed product" is shrinking by the day. Community > Code: The most powerful breakthroughs happen when we step away from the screen and start talking to each other. NYC is the place to be: The density of talent here is unmatched. Huge thanks to the organizers Aishwarya Srinivasan Arvind Narayanamurthy Aditya Suresh and everyone I chatted with today. If we met, letā€™s keep the conversation going! šŸ‘‡ Special thanks to Anna Mai from Pinecone for the incredible insights. #AI #NYCTech #ArtificialIntelligence #Networking #Innovation #GenerativeAI #Newyork

SV
Suman Vigrahala
I spent a few hours at the Pinecone x The Gen Academy, AI Meetup in NYC - a gathering of builders, founders, and AI-curious folks! Anna Mai (Pinecone): for the in-depth presentation on the Pinecone Nexus model, a Knowledge Engine for Agents, and walking through a live demo. Seeing it in motion, not just on slides, is what made the capabilities click for me. Truly appreciated the time and effort - a special thank you! Arvind Narayanamurthy - listening to you go technical on RAG and Evals, then getting a quick chat in person, was a highlight. The way you frame AI concepts through practical, working models is something I'm excited to learn! Aditya Suresh - thank you for the friendly and generous conversation on Product Management for AI platforms, and for breaking down what it looks like to build AI infrastructure. Didn't realize how much I needed those insights until I had them. Looking forward to staying in touch and meeting you in the future! Aishwarya Srinivasan - I'm a long-time admirer of your talent and consistent hard work, and your passion for educating people about AI at every level. Your RAG Explained video is one of my favorites - it sparked my interest and motivated me to keep learning to become AI-native. Meeting you in person and hearing you teach in real time was a fan moment! The AI interactive quiz was cool & fun, and walking out with copies of Patterns for Building AI Agents and Principles of Building AI Agents (Sam Bhagwat, Mastra.ai) was the key takeaway! For me, days like this are part of a deliberate shift - evolving to be AI-native. Getting fluent in RAG, LLMs, evals, and security, and genuinely excited to integrate everything I'm learning with my Product Management principles and experience on AI platforms! #AI #ProductManagement #RAG #AINative
I spent a few hours at the Pinecone x The Gen Academy, AI Meetup in NYC - a gathering of builders, founders, and AI-curious folks! Anna Mai (Pinecone): for the in-depth presentation on the Pinecone Nexus model, a Knowledge Engine for Agents, and walking through a live demo. Seeing it in motion, not just on slides, is what made the capabilities click for me. Truly appreciated the time and effort - a special thank you! Arvind Narayanamurthy - listening to you go technical on RAG and Evals, then getting a quick chat in person, was a highlight. The way you frame AI concepts through practical, working models is something I'm excited to learn! Aditya Suresh - thank you for the friendly and generous conversation on Product Management for AI platforms, and for breaking down what it looks like to build AI infrastructure. Didn't realize how much I needed those insights until I had them. Looking forward to staying in touch and meeting you in the future! Aishwarya Srinivasan - I'm a long-time admirer of your talent and consistent hard work, and your passion for educating people about AI at every level. Your RAG Explained video is one of my favorites - it sparked my interest and motivated me to keep learning to become AI-native. Meeting you in person and hearing you teach in real time was a fan moment! The AI interactive quiz was cool & fun, and walking out with copies of Patterns for Building AI Agents and Principles of Building AI Agents (Sam Bhagwat, Mastra.ai) was the key takeaway! For me, days like this are part of a deliberate shift - evolving to be AI-native. Getting fluent in RAG, LLMs, evals, and security, and genuinely excited to integrate everything I'm learning with my Product Management principles and experience on AI platforms! #AI #ProductManagement #RAG #AINative

Atharv Patole
That moment when you finally understand what everyone's talking about? Just had that at an AI event in NYC Attended this exclusive AI meetup at Pinecone HQ, and the vibe was immaculate ā€” no corporate fluff, just actual builders and engineers talking about what they're shipping. The conversations went beyond theory; people were sharing how they're actually using AI in their work, their strategies for handling tokens efficiently, and how they're navigating privacy and security concerns. That's the real stuff that matters. The panel with Aishwarya Srinivasan and Arvind Narayanamurthy broke down how to build production-ready AI systems that actually work in the real world. We went deep into RAG systems, agentic workflows, and the practical challenges people face daily. Watching PMs, engineers, and builders exchange ideas and real use cases was genuinely valuable. Honestly, I felt like I was lagging on the whole AI thing, but being here flipped a switch. Now I get it ā€” it's not about chasing every update, it's about building, shipping, and getting actual visibility on what works. That's the move.
That moment when you finally understand what everyone's talking about? Just had that at an AI event in NYC Attended this exclusive AI meetup at Pinecone HQ, and the vibe was immaculate ā€” no corporate fluff, just actual builders and engineers talking about what they're shipping. The conversations went beyond theory; people were sharing how they're actually using AI in their work, their strategies for handling tokens efficiently, and how they're navigating privacy and security concerns. That's the real stuff that matters. The panel with Aishwarya Srinivasan and Arvind Narayanamurthy broke down how to build production-ready AI systems that actually work in the real world. We went deep into RAG systems, agentic workflows, and the practical challenges people face daily. Watching PMs, engineers, and builders exchange ideas and real use cases was genuinely valuable. Honestly, I felt like I was lagging on the whole AI thing, but being here flipped a switch. Now I get it ā€” it's not about chasing every update, it's about building, shipping, and getting actual visibility on what works. That's the move.

khyathi koteru
Excited to share that Iā€™ve successfully completed the Post Graduate Program in Artificial Intelligence and Machine Learning: Business Applications from The University of Texas at Austin ā€“ McCombs School of Business. This journey gave me hands-on exposure to machine learning, deep learning, NLP, predictive analytics, Generative AI, and Retrieval-Augmented Generation (RAG) architectures, along with building and evaluating AI-driven business solutions. It was exciting to learn how AI/ML can be applied to create more intelligent, scalable, and user-focused products. Iā€™m especially interested in the growing space of GenAI applications, RAG-based systems, and AI-powered experiences. Balancing full-time work, learning, and family responsibilities made this journey both challenging and rewarding, and Iā€™m incredibly grateful for the support system that helped me through it. A special thank you to my manager, Shawn Hooley, and my team at work for their encouragement and support in fostering a culture of continuous learning, growth, and exploration of new technologies. The learning journey continues, and Iā€™m excited to next explore Agentic AI, autonomous systems, and advanced GenAI workflows through NVIDIA partnered Agentic AI program by Aishwarya Srinivasan and Arvind Narayanamurthy. As a working mom, one of my biggest motivations was to set an example for my kids that learning and growth never stop. Sharing this milestone on Motherā€™s Day makes it even more meaningful. Happy Motherā€™s Day to all the moms balancing dreams, responsibilities, careers, and family every single day. #ArtificialIntelligence #MachineLearning #GenerativeAI #RAG #LLM #NLP #WomenInTech #WorkingMom #UTAustin #MothersDay
Excited to share that Iā€™ve successfully completed the Post Graduate Program in Artificial Intelligence and Machine Learning: Business Applications from The University of Texas at Austin ā€“ McCombs School of Business. This journey gave me hands-on exposure to machine learning, deep learning, NLP, predictive analytics, Generative AI, and Retrieval-Augmented Generation (RAG) architectures, along with building and evaluating AI-driven business solutions. It was exciting to learn how AI/ML can be applied to create more intelligent, scalable, and user-focused products. Iā€™m especially interested in the growing space of GenAI applications, RAG-based systems, and AI-powered experiences. Balancing full-time work, learning, and family responsibilities made this journey both challenging and rewarding, and Iā€™m incredibly grateful for the support system that helped me through it. A special thank you to my manager, Shawn Hooley, and my team at work for their encouragement and support in fostering a culture of continuous learning, growth, and exploration of new technologies. The learning journey continues, and Iā€™m excited to next explore Agentic AI, autonomous systems, and advanced GenAI workflows through NVIDIA partnered Agentic AI program by Aishwarya Srinivasan and Arvind Narayanamurthy. As a working mom, one of my biggest motivations was to set an example for my kids that learning and growth never stop. Sharing this milestone on Motherā€™s Day makes it even more meaningful. Happy Motherā€™s Day to all the moms balancing dreams, responsibilities, careers, and family every single day. #ArtificialIntelligence #MachineLearning #GenerativeAI #RAG #LLM #NLP #WomenInTech #WorkingMom #UTAustin #MothersDay

Nithya Ranganathan
š…š«šØš¦ š…š„šØš©š©š² šƒš¢š¬š¤š¬ š­šØ šŒš®š„š­š¢-š€š šžš§š­ š’š²š¬š­šžš¦š¬: š“š”š¢š¬ š’š”š¢šŸš­ š…šžšžš„š¬ šƒš¢šŸšŸšžš«šžš§š­. š–£š–¾š–¼š–ŗš–½š–¾š—Œ š–ŗš—€š—ˆ, š–Ø š–¼š–ŗš—‹š—‹š—‚š–¾š–½ š–ŗ š–暝—…š—ˆš—‰š—‰š—’ š–½š—‚š—Œš—„ š–ŗš–¼š—‹š—ˆš—Œš—Œ š—š—ˆš–¶š—‡ š—‚š—‡ š–ŗš—‡ š–ŗš—Žš–™š—ˆš—‹š—‚š–¼š—„š—Œš—š–ŗš—. š–©š—Žš—Œš— š–³š—ˆ š—‚š—‡š—Œš—š–ŗš—…š—… š–ŗ š—Œš—ˆš–暝–¾ š–ŗš—‹š–¾ š–ŗ š–ŗš—š—‡ā€™š— š–ŗ š–¼š–«š—‚š–¼š—„. š–£š–¾š—‰š—…š—ˆš—’š—†š–¾š—‡š— š—š—®š–¾ š–ŗ š–©š—ˆš—Žš—‹š—‡š–¾š–ø. š“š”šžš§ š­š”šž š°šØš«š„š šœš”ššš§š šžš. š–²š–ŗš–ŗš–². š–¢š—…š—ˆš—Žš–½. š–¶š—ˆš—‹š—„š–暝—…š—ˆš—š—Œ. š–²š—ˆš—™š–³š–¶š–ŗš—‹š–¾ š–»š–¾š–¼š–ŗš—†š–¾ š–暝–ŗš—Œš—š–¾š—‹ š–ŗš—‡š–½ š–²š–¼š–ŗš—…š–ŗš–»š—…š–¾. šš®š­ š­š”š¢š¬ š­š¢š¦šž, š­š”šž š„šžššš© š¢š¬ ššš«šœš”š¢š­šžšœš­š®š«ššš„. š–©š—Žš—Œš— š–  š—’š–¾š–ŗš—‹ š—ˆš—‹ š—š–‚š—ˆ š–ŗš—€š—ˆ, š—š–¾ š–œš–¾š—‹š–¾ š–暝—ˆš–¼š—Žš—Œš–¾š–½ š—ˆš—‡ š—‚š—‡š–½š—‚š—š—‚š–½š—Žš–ŗš—… š—‰š—‹š—ˆš—†š—‰š—š—Œ š–ŗš—‡š–½ š–«š–«š–¬ š—ˆš—Žš—š—‰š—Žš—š—Œ. š–³š—š–¾š—‡ š–¼š–ŗš—†š–¾ š–œš–¾š—‹š–½š–暝—…š—ˆš— š—ˆš—‹š—–š—š–¾š—Œš—š—‹š–ŗš—š–¾š—ˆš—‡ā€”š—†š—ˆš–µš—‚š—‡š—€ š–»š–¾š—’š—ˆš—‡š–½ š–¼š—š–ŗš— š—š—ˆ š–²š—š—‹š—Žš–¼š—š—Žš—‹š–¾š–½ š–ŗš—Žš—š—ˆš—†š–ŗš—š—‚š—ˆš—‡. š–Øš–³ š—š–ŗš—Œ š—‰š—‹š—ˆš—€š—‹š–¾š—Œš–¾š–½, š–»š—Žš— š—‚š— š—š–ŗš—Œ š—‹š—‚š—€š—‚š–½. š–¶š—‚š—š— š–¬š—’ š–»š–ŗš–¼š—„š—€š—‹š—ˆš—Žš—‡š–½ š—‚š—‡ š–¤š–³š–« š–ŗš—‡š–½ š—‰š—‹š—ˆš–¼š–¾š—Œš—Œ š—†š–ŗš—‰š—‰š—‚š—‡š—€, šˆ šœšØš®š„š š›š®š¢š„š š­š”šž š„šØš š¢šœ. š–”š–“š— š–Øš— š–œš–ŗš—Œ š–²š—š—‚š—…š—… š–²š—ˆš—†š–¾ š—š–¾š—‹š—„. š–¤š–µš–¾š—‹š—’ š–»š—‹š–ŗš—‡š–¼š–§ š—š–ŗš–½ š–³š—ˆ š–»š–¾ š–¼š—ˆš–½š–¾š–½. š–¤š—š–¾š—‹š—’ š–¾š–½š—€š–¾ š–¼š–ŗš—Œš–¾ š—‰š—‹š–¾š–½š—‚š–¼š—š–¾š–½. š–Øš–³ š—š–ŗš—Œ š–ŗš–šš–³š—ˆš—†š–ŗš–³š—‚š—ˆš—‡, š–»š—Žš— š—‚š— š–œš–ŗš—Œš—‡'š— š–øš–¾š–¹ š—’š–¾š–³ "š—‚š—‡š–³š–¾š—…š–«š—‚š–¾š—‡š–¾š–½". š“š”šžš§ šœššš¦šž š­š”šž š¬š”š¢šŸš­ š­šØ š€š šžš§š­š¬. š–Ø š—†š—ˆš—š–¾š–½ š–暝—‹š—ˆš—† š–»š—Žš—‚š—…š–½š—‚š—‡š—€ š–²š—š–ŗš—š—‚š–¼ š–暝—…š—ˆš—š—Œ š—š—ˆ š—ˆš—‹š–¼š—š–¾š–¾š–²š—š—‹š–ŗš–¾š—…š—‚š—‡š—€ š—Œš—’š–²š—š–¾š—†š—Œ š—š—‚š–³š— š‘Ŗš’š’‚š’–š’…š’† š‘Ŗš’š’˜š’š’“š’Œ š–ŗš—‡š–½ š‘Ŗš’“š’†š’˜š‘Øš‘°. š‚š«šžš°š€šˆ, š—‚š—‡ š—‰š–ŗš—‹š—š—‚š–¼š—Žš—…š–ŗš—‹, š—‚š— š–¼š—š–ŗš—‡š–¼š–¾š–½ š—š—š–¾ š—€š–ŗš—†š–¾ š–暝—ˆš—‹ š—†š–¾. š–Øš—‡š–²š—š–¾š–ŗš–½ š—ˆš–æ š–½š–¾š–暝—‚š—‡š—‚š—‡š—€ š–¾š—š–¾š—‹š–ø š–¬š—‚š—†š—‹š—ˆ-š–²š—š–¾š—‰, šˆ š¬š­ššš«š­šžš ššžš¬šœš«š¢š›š¢š§š  š­š”šž š¢š§š­šžš§š­ ššš§š š«šØš„šžš¬. š€š§š ššØšØš¦ ā€“ š–¢š—‹š—ˆš— š– š–Ø built a multi-agent system end-to-end. š–³š—š–¾ š—Œš–¾š—Œš—š–¾š—† š–½š–¾š—Œš—‚š–Œš—‡š—Œ š—š—š–¾ š—‰š–ŗš—š—. š– š—€š–¾š—‡š–³š—Œ š–¼š—ˆš—…š—…š–ŗš–»š—ˆš–—š–ŗš–³š–¾. š–¤š–·š–¾š–¼š–¼š—Žš—š—‚š—ˆš—‡ š—š–ŗš—‰š—‰š–¾š—‡š—Œ š–ŗš— š—Œš–¼š–ŗš—…š–¾. š–Øā€™š–µš–¾ š—Œš—š—‚š–暝—š–¾š–½ š–暝—‹š—ˆš—† š–»š—Žš—‚š—…š–½š–¾š–½ š—š—š–¾ š—š—‹š–ŗš–¼š—„š—Œ š—š—ˆ šš¢š«šžšœš­š¢š§š  š­š”šž š¢š§š­šžš„š„š¢š šžš§šœšž. š€š¬ ššš§ š€šˆ š©š«šššœš­š¢š­š¢šØš§šžš«, šˆā€™šÆšž š„šžššš«š§šžš š­š”ššš­ š­š”šž š«šžššš„ š›šØš­š­š„šžš§šžšœš¤ š¢š¬š§'š­ š¢š§š­šžš„š„š¢š šžš§šœšžā€”š¢š­ā€™š¬ šžšŸšŸš¢šœš¢šžš§šœš². š–³š—ˆš—„š–¾š—‡š—Œ š–ŗš—‹š–¾ š—š—š–¾ š—‡š–¾š— "š—€š–ŗš—Œ š—†š—‚š—…š–¾š–ŗš—€š–¾" š—ˆš–æ š—š—š–¾ š–¾š—‡š—š–¾š—‹š—‰š—‹š—‚š–²š–¾. š–„š—ˆš—‹ š—š—ˆš—„š–¾š—‡ š—ˆš—‰š—š—‚š—†š—‚š–暝–ŗš—š—‚š—ˆš—‡, šˆā€™šÆšž š›š®š¢š„š­ ššš§ ššš©š© šØš§ šš®š›š›š„šž š€šˆ ā€“ š§šØ šœšØššž š°š«š¢š­š­šžš§. š‘¾š’† š’‚š’“š’†š’ā€™š’• š’‹š’–š’”š’• š’–š’”š’Šš’š’ˆ š’”š’š’‡š’•š’˜š’‚š’“š’† š’‚š’š’šš’Žš’š’“š’†. š‘¾š’† š’‚š’“š’† š’„š’-š’„š’“š’†š’‚š’•š’Šš’š’ˆ š’Šš’•. #AI #GenerativeAI #CrewAI #Claude #BubbleAI #RevenueOps #AppliedAI #AIOrchestration ššØš­šž: š˜š˜§ š˜ŗš˜°š˜¶ š˜øš˜¢š˜Æš˜µ š˜µš˜° š˜­š˜¦š˜¢š˜³š˜Æ š˜Šš˜­š˜¢š˜¶š˜„š˜¦ š˜Šš˜°š˜øš˜°š˜³š˜¬, š˜ š˜©š˜Ŗš˜Øš˜©š˜­š˜ŗ š˜³š˜¦š˜¤š˜°š˜®š˜®š˜¦š˜Æš˜„ š˜øš˜¢š˜µš˜¤š˜©š˜Ŗš˜Æš˜Ø š˜µš˜©š˜Ŗš˜“ š˜µš˜¶š˜µš˜°š˜³š˜Ŗš˜¢š˜­ š˜£š˜ŗ Aishwarya Srinivasan - š„š¢š§š¤ š¢š§ šœšØš¦š¦šžš§š­ š›šžš„šØš°!
š…š«šØš¦ š…š„šØš©š©š² šƒš¢š¬š¤š¬ š­šØ šŒš®š„š­š¢-š€š šžš§š­ š’š²š¬š­šžš¦š¬: š“š”š¢š¬ š’š”š¢šŸš­ š…šžšžš„š¬ šƒš¢šŸšŸšžš«šžš§š­. š–£š–¾š–¼š–ŗš–½š–¾š—Œ š–ŗš—€š—ˆ, š–Ø š–¼š–ŗš—‹š—‹š—‚š–¾š–½ š–ŗ š–暝—…š—ˆš—‰š—‰š—’ š–½š—‚š—Œš—„ š–ŗš–¼š—‹š—ˆš—Œš—Œ š—š—ˆš–¶š—‡ š—‚š—‡ š–ŗš—‡ š–ŗš—Žš–™š—ˆš—‹š—‚š–¼š—„š—Œš—š–ŗš—. š–©š—Žš—Œš— š–³š—ˆ š—‚š—‡š—Œš—š–ŗš—…š—… š–ŗ š—Œš—ˆš–暝–¾ š–ŗš—‹š–¾ š–ŗ š–ŗš—š—‡ā€™š— š–ŗ š–¼š–«š—‚š–¼š—„. š–£š–¾š—‰š—…š—ˆš—’š—†š–¾š—‡š— š—š—®š–¾ š–ŗ š–©š—ˆš—Žš—‹š—‡š–¾š–ø. š“š”šžš§ š­š”šž š°šØš«š„š šœš”ššš§š šžš. š–²š–ŗš–ŗš–². š–¢š—…š—ˆš—Žš–½. š–¶š—ˆš—‹š—„š–暝—…š—ˆš—š—Œ. š–²š—ˆš—™š–³š–¶š–ŗš—‹š–¾ š–»š–¾š–¼š–ŗš—†š–¾ š–暝–ŗš—Œš—š–¾š—‹ š–ŗš—‡š–½ š–²š–¼š–ŗš—…š–ŗš–»š—…š–¾. šš®š­ š­š”š¢š¬ š­š¢š¦šž, š­š”šž š„šžššš© š¢š¬ ššš«šœš”š¢š­šžšœš­š®š«ššš„. š–©š—Žš—Œš— š–  š—’š–¾š–ŗš—‹ š—ˆš—‹ š—š–‚š—ˆ š–ŗš—€š—ˆ, š—š–¾ š–œš–¾š—‹š–¾ š–暝—ˆš–¼š—Žš—Œš–¾š–½ š—ˆš—‡ š—‚š—‡š–½š—‚š—š—‚š–½š—Žš–ŗš—… š—‰š—‹š—ˆš—†š—‰š—š—Œ š–ŗš—‡š–½ š–«š–«š–¬ š—ˆš—Žš—š—‰š—Žš—š—Œ. š–³š—š–¾š—‡ š–¼š–ŗš—†š–¾ š–œš–¾š—‹š–½š–暝—…š—ˆš— š—ˆš—‹š—–š—š–¾š—Œš—š—‹š–ŗš—š–¾š—ˆš—‡ā€”š—†š—ˆš–µš—‚š—‡š—€ š–»š–¾š—’š—ˆš—‡š–½ š–¼š—š–ŗš— š—š—ˆ š–²š—š—‹š—Žš–¼š—š—Žš—‹š–¾š–½ š–ŗš—Žš—š—ˆš—†š–ŗš—š—‚š—ˆš—‡. š–Øš–³ š—š–ŗš—Œ š—‰š—‹š—ˆš—€š—‹š–¾š—Œš–¾š–½, š–»š—Žš— š—‚š— š—š–ŗš—Œ š—‹š—‚š—€š—‚š–½. š–¶š—‚š—š— š–¬š—’ š–»š–ŗš–¼š—„š—€š—‹š—ˆš—Žš—‡š–½ š—‚š—‡ š–¤š–³š–« š–ŗš—‡š–½ š—‰š—‹š—ˆš–¼š–¾š—Œš—Œ š—†š–ŗš—‰š—‰š—‚š—‡š—€, šˆ šœšØš®š„š š›š®š¢š„š š­š”šž š„šØš š¢šœ. š–”š–“š— š–Øš— š–œš–ŗš—Œ š–²š—š—‚š—…š—… š–²š—ˆš—†š–¾ š—š–¾š—‹š—„. š–¤š–µš–¾š—‹š—’ š–»š—‹š–ŗš—‡š–¼š–§ š—š–ŗš–½ š–³š—ˆ š–»š–¾ š–¼š—ˆš–½š–¾š–½. š–¤š—š–¾š—‹š—’ š–¾š–½š—€š–¾ š–¼š–ŗš—Œš–¾ š—‰š—‹š–¾š–½š—‚š–¼š—š–¾š–½. š–Øš–³ š—š–ŗš—Œ š–ŗš–šš–³š—ˆš—†š–ŗš–³š—‚š—ˆš—‡, š–»š—Žš— š—‚š— š–œš–ŗš—Œš—‡'š— š–øš–¾š–¹ š—’š–¾š–³ "š—‚š—‡š–³š–¾š—…š–«š—‚š–¾š—‡š–¾š–½". š“š”šžš§ šœššš¦šž š­š”šž š¬š”š¢šŸš­ š­šØ š€š šžš§š­š¬. š–Ø š—†š—ˆš—š–¾š–½ š–暝—‹š—ˆš—† š–»š—Žš—‚š—…š–½š—‚š—‡š—€ š–²š—š–ŗš—š—‚š–¼ š–暝—…š—ˆš—š—Œ š—š—ˆ š—ˆš—‹š–¼š—š–¾š–¾š–²š—š—‹š–ŗš–¾š—…š—‚š—‡š—€ š—Œš—’š–²š—š–¾š—†š—Œ š—š—‚š–³š— š‘Ŗš’š’‚š’–š’…š’† š‘Ŗš’š’˜š’š’“š’Œ š–ŗš—‡š–½ š‘Ŗš’“š’†š’˜š‘Øš‘°. š‚š«šžš°š€šˆ, š—‚š—‡ š—‰š–ŗš—‹š—š—‚š–¼š—Žš—…š–ŗš—‹, š—‚š— š–¼š—š–ŗš—‡š–¼š–¾š–½ š—š—š–¾ š—€š–ŗš—†š–¾ š–暝—ˆš—‹ š—†š–¾. š–Øš—‡š–²š—š–¾š–ŗš–½ š—ˆš–æ š–½š–¾š–暝—‚š—‡š—‚š—‡š—€ š–¾š—š–¾š—‹š–ø š–¬š—‚š—†š—‹š—ˆ-š–²š—š–¾š—‰, šˆ š¬š­ššš«š­šžš ššžš¬šœš«š¢š›š¢š§š  š­š”šž š¢š§š­šžš§š­ ššš§š š«šØš„šžš¬. š€š§š ššØšØš¦ ā€“ š–¢š—‹š—ˆš— š– š–Ø built a multi-agent system end-to-end. š–³š—š–¾ š—Œš–¾š—Œš—š–¾š—† š–½š–¾š—Œš—‚š–Œš—‡š—Œ š—š—š–¾ š—‰š–ŗš—š—. š– š—€š–¾š—‡š–³š—Œ š–¼š—ˆš—…š—…š–ŗš–»š—ˆš–—š–ŗš–³š–¾. š–¤š–·š–¾š–¼š–¼š—Žš—š—‚š—ˆš—‡ š—š–ŗš—‰š—‰š–¾š—‡š—Œ š–ŗš— š—Œš–¼š–ŗš—…š–¾. š–Øā€™š–µš–¾ š—Œš—š—‚š–暝—š–¾š–½ š–暝—‹š—ˆš—† š–»š—Žš—‚š—…š–½š–¾š–½ š—š—š–¾ š—š—‹š–ŗš–¼š—„š—Œ š—š—ˆ šš¢š«šžšœš­š¢š§š  š­š”šž š¢š§š­šžš„š„š¢š šžš§šœšž. š€š¬ ššš§ š€šˆ š©š«šššœš­š¢š­š¢šØš§šžš«, šˆā€™šÆšž š„šžššš«š§šžš š­š”ššš­ š­š”šž š«šžššš„ š›šØš­š­š„šžš§šžšœš¤ š¢š¬š§'š­ š¢š§š­šžš„š„š¢š šžš§šœšžā€”š¢š­ā€™š¬ šžšŸšŸš¢šœš¢šžš§šœš². š–³š—ˆš—„š–¾š—‡š—Œ š–ŗš—‹š–¾ š—š—š–¾ š—‡š–¾š— "š—€š–ŗš—Œ š—†š—‚š—…š–¾š–ŗš—€š–¾" š—ˆš–æ š—š—š–¾ š–¾š—‡š—š–¾š—‹š—‰š—‹š—‚š–²š–¾. š–„š—ˆš—‹ š—š—ˆš—„š–¾š—‡ š—ˆš—‰š—š—‚š—†š—‚š–暝–ŗš—š—‚š—ˆš—‡, šˆā€™šÆšž š›š®š¢š„š­ ššš§ ššš©š© šØš§ šš®š›š›š„šž š€šˆ ā€“ š§šØ šœšØššž š°š«š¢š­š­šžš§. š‘¾š’† š’‚š’“š’†š’ā€™š’• š’‹š’–š’”š’• š’–š’”š’Šš’š’ˆ š’”š’š’‡š’•š’˜š’‚š’“š’† š’‚š’š’šš’Žš’š’“š’†. š‘¾š’† š’‚š’“š’† š’„š’-š’„š’“š’†š’‚š’•š’Šš’š’ˆ š’Šš’•. #AI #GenerativeAI #CrewAI #Claude #BubbleAI #RevenueOps #AppliedAI #AIOrchestration ššØš­šž: š˜š˜§ š˜ŗš˜°š˜¶ š˜øš˜¢š˜Æš˜µ š˜µš˜° š˜­š˜¦š˜¢š˜³š˜Æ š˜Šš˜­š˜¢š˜¶š˜„š˜¦ š˜Šš˜°š˜øš˜°š˜³š˜¬, š˜ š˜©š˜Ŗš˜Øš˜©š˜­š˜ŗ š˜³š˜¦š˜¤š˜°š˜®š˜®š˜¦š˜Æš˜„ š˜øš˜¢š˜µš˜¤š˜©š˜Ŗš˜Æš˜Ø š˜µš˜©š˜Ŗš˜“ š˜µš˜¶š˜µš˜°š˜³š˜Ŗš˜¢š˜­ š˜£š˜ŗ Aishwarya Srinivasan - š„š¢š§š¤ š¢š§ šœšØš¦š¦šžš§š­ š›šžš„šØš°!

Prathyusha D.
Multimodal AI is moving us beyond text-based interactions. This article from The Gen Academy highlights how AI is rapidly evolving to interpret and respond across speech, images, and video, not just written input: lnkd.in/gE9GD_AW In practice, Iā€™m already seeing this shift, using the speech option in ChatGPT consistently yields more context-rich, natural responses compared to text alone. This isnā€™t incremental progress. Itā€™s a change in how humans and systems will interact - more intuitive, more contextual, and closer to real-world communication. Worth paying attention to, as we design the next generation of products and experiences. Thank you Aishwarya Srinivasan for sharing your expertise to the rest of the world, really appreciate your work!
Multimodal AI is moving us beyond text-based interactions. This article from The Gen Academy highlights how AI is rapidly evolving to interpret and respond across speech, images, and video, not just written input: lnkd.in/gE9GD_AW In practice, Iā€™m already seeing this shift, using the speech option in ChatGPT consistently yields more context-rich, natural responses compared to text alone. This isnā€™t incremental progress. Itā€™s a change in how humans and systems will interact - more intuitive, more contextual, and closer to real-world communication. Worth paying attention to, as we design the next generation of products and experiences. Thank you Aishwarya Srinivasan for sharing your expertise to the rest of the world, really appreciate your work!

Joelle Guerra
ā˜•Most AI training creates spectators. This one creates builders. I met the founders of The Gen Academy last week in San Francisco and their conviction is hard to argue with. In 2025, they kept seeing the same thing: ambitious professionals surrounded by AI toolsā€¦ who didn't know how to actually build with them. Tutorials taught syntax. Conferences sold inspiration. Almost no one taught the messy middle... how to design, ship, and trust real AI systems. So they built the program they wished existed. For builders. Not spectators. What struck me most: -> Learn by building : every workshop ships a working project. Notes are nice, repos are better. -> Uses cases and the ecosystem as the classroom : learners work with the real building blocks from different tech partners. No toy demos in a notebook. -> A global community from day one : Cohort 1 sold out, with professionals from tech giants, financial institutions, consulting firms and retail leadersā€¦ across more than ten countries. And their take on go to market goes well beyond education: "You don't sell to developers. You earn them." What made this encounter particularly inspiring for me: rarely do you find coaching, mentoring, functional expertise and deep technical knowledge all in the same room. Most training programs offer one or the other. Here, it's all of it and that very smart combination hits differently ! It changes the questions you ask. And the ones you bring back home. In a world where AI tools shift every week, the real question is no longer "who has the best model" - it's who is actually training people to build with it. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for the conversation. I wish the best to your GenAcademy šŸ˜€ For more info : lnkd.in/eEcj_TfB *** Pour quelques jours, les Pauses IA passent Ć  l'anglais :) mĆŖme format, mĆŖme heure ou presque, autre langue. Une immersion de terrain, pour ramener l'essentiel aux Ć©quipes et aux projets.*** #AI #AIEducation #GenAI #BuildersNotSpectators
ā˜•Most AI training creates spectators. This one creates builders. I met the founders of The Gen Academy last week in San Francisco and their conviction is hard to argue with. In 2025, they kept seeing the same thing: ambitious professionals surrounded by AI toolsā€¦ who didn't know how to actually build with them. Tutorials taught syntax. Conferences sold inspiration. Almost no one taught the messy middle... how to design, ship, and trust real AI systems. So they built the program they wished existed. For builders. Not spectators. What struck me most: -> Learn by building : every workshop ships a working project. Notes are nice, repos are better. -> Uses cases and the ecosystem as the classroom : learners work with the real building blocks from different tech partners. No toy demos in a notebook. -> A global community from day one : Cohort 1 sold out, with professionals from tech giants, financial institutions, consulting firms and retail leadersā€¦ across more than ten countries. And their take on go to market goes well beyond education: "You don't sell to developers. You earn them." What made this encounter particularly inspiring for me: rarely do you find coaching, mentoring, functional expertise and deep technical knowledge all in the same room. Most training programs offer one or the other. Here, it's all of it and that very smart combination hits differently ! It changes the questions you ask. And the ones you bring back home. In a world where AI tools shift every week, the real question is no longer "who has the best model" - it's who is actually training people to build with it. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for the conversation. I wish the best to your GenAcademy šŸ˜€ For more info : lnkd.in/eEcj_TfB *** Pour quelques jours, les Pauses IA passent Ć  l'anglais :) mĆŖme format, mĆŖme heure ou presque, autre langue. Une immersion de terrain, pour ramener l'essentiel aux Ć©quipes et aux projets.*** #AI #AIEducation #GenAI #BuildersNotSpectators

Tanmay Pathak
I'll be honest ā€” I went into today's session 'How AI Changes Design: A Practical Look' thinking I already 'got' how AI fits into design. I was wrong. 'How AI Changes Design: A Practical Look' with Aishwarya Srinivasan, Arvind Narayanamurthy and Aishwarya Rao reframed how I think about the AI products. The part that hit hardest? Aishwarya Rao's point that AI isn't just a faster way to do the same job ā€” it changes what the job *is*. That's a different conversation than most people are having. Whether you're in design, product, or engineering ā€” if AI touches your work in any way, the frameworks from this session are worth your time. Huge appreciation for industry experts who bring genuine depth to the conversation ā€” this is the kind of session that actually shapes how you think and build. šŸ’” šŸ™Œ
I'll be honest ā€” I went into today's session 'How AI Changes Design: A Practical Look' thinking I already 'got' how AI fits into design. I was wrong. 'How AI Changes Design: A Practical Look' with Aishwarya Srinivasan, Arvind Narayanamurthy and Aishwarya Rao reframed how I think about the AI products. The part that hit hardest? Aishwarya Rao's point that AI isn't just a faster way to do the same job ā€” it changes what the job *is*. That's a different conversation than most people are having. Whether you're in design, product, or engineering ā€” if AI touches your work in any way, the frameworks from this session are worth your time. Huge appreciation for industry experts who bring genuine depth to the conversation ā€” this is the kind of session that actually shapes how you think and build. šŸ’” šŸ™Œ

Mujahid Mirza

Driving Data Strategy & Insights | BI, AI & Data Engineering

Last weekend, I attended a session on š— š—®š˜€š˜š—²š—暝—¶š—»š—“ š—–š—¹š—®š˜‚š—±š—² š—–š—¼š˜„š—¼š—暝—ø hosted by Aishwarya Srinivasan and Arvind Narayanamurthy, and it shifted how I think about data exploration and problem-solving. š—žš—²š˜† š—Ÿš—²š—®š—暝—»š—¶š—»š—“š˜€: ā€¢ š—œš—»š˜€š˜š—®š—»š˜ š—”š—Ÿ š—˜š˜…š—½š—¹š—¼š—暝—®š˜š—¶š—¼š—» - Natural language forces clearer problem definition before diving into implementation (No Sql) ā€¢ š—„š—®š—½š—¶š—± & š—Ÿš—œš—©š—˜ š—£š—暝—¼š˜š—¼š˜š˜†š—½š—¶š—»š—“ - Accelerated hypothesis testing that I'm now applying to my BI workflows  ā€¢ š—–š—¼š—¹š—¹š—®š—Æš—¼š—暝—®š˜š—¶š˜ƒš—² š——š—®š˜š—® š— š—¼š—±š—²š—¹š—¶š—»š—“ - Breaking down schemas through conversation reveals insights I'd miss traditionally  ā€¢ š—„š—²š˜‚š˜€š—®š—Æš—¹š—² š——š—²š˜€š—¶š—“š—» š—£š—®š˜š˜š—²š—暝—»š˜€ - The markdown-based approach to documentation is game-changing To validate these learnings, I built a FIFA 2022 analytics dashboard using Claude Cowork as a sandbox to test concepts before applying them to enterprise dashboards. šŸ“Š š——š—®š˜€š—µš—Æš—¼š—®š—暝—±: lnkd.in/g6m9PVzQ  šŸ”— š—šš—¶š˜š—›š˜‚š—Æ: lnkd.in/gteareCy Are you experimenting with these shifts? Would love your thoughts.
Last weekend, I attended a session on š— š—®š˜€š˜š—²š—暝—¶š—»š—“ š—–š—¹š—®š˜‚š—±š—² š—–š—¼š˜„š—¼š—暝—ø hosted by Aishwarya Srinivasan and Arvind Narayanamurthy, and it shifted how I think about data exploration and problem-solving. š—žš—²š˜† š—Ÿš—²š—®š—暝—»š—¶š—»š—“š˜€: ā€¢ š—œš—»š˜€š˜š—®š—»š˜ š—”š—Ÿ š—˜š˜…š—½š—¹š—¼š—暝—®š˜š—¶š—¼š—» - Natural language forces clearer problem definition before diving into implementation (No Sql) ā€¢ š—„š—®š—½š—¶š—± & š—Ÿš—œš—©š—˜ š—£š—暝—¼š˜š—¼š˜š˜†š—½š—¶š—»š—“ - Accelerated hypothesis testing that I'm now applying to my BI workflows  ā€¢ š—–š—¼š—¹š—¹š—®š—Æš—¼š—暝—®š˜š—¶š˜ƒš—² š——š—®š˜š—® š— š—¼š—±š—²š—¹š—¶š—»š—“ - Breaking down schemas through conversation reveals insights I'd miss traditionally  ā€¢ š—„š—²š˜‚š˜€š—®š—Æš—¹š—² š——š—²š˜€š—¶š—“š—» š—£š—®š˜š˜š—²š—暝—»š˜€ - The markdown-based approach to documentation is game-changing To validate these learnings, I built a FIFA 2022 analytics dashboard using Claude Cowork as a sandbox to test concepts before applying them to enterprise dashboards. šŸ“Š š——š—®š˜€š—µš—Æš—¼š—®š—暝—±: lnkd.in/g6m9PVzQ  šŸ”— š—šš—¶š˜š—›š˜‚š—Æ: lnkd.in/gteareCy Are you experimenting with these shifts? Would love your thoughts.

Anand Sridharan

Service Delivery Transformation | Leading Intelligent Ops | Data & AI

From Text to SQL: How SQL AI Agents are Transforming Data Access Wrapped up an incredible session on š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ with Rami Krispin, where he shared his rich experience and production ready play-book. Encapsulated - Top 5 pointers (provided AI generated image for illustration only) //Context is King, Not the Model You don't need the fanciest LLM to build powerful SQL agents. A thoughtfully engineered prompt with rich context (schema details, categorical values, business logic) can outperform expensive models. It's about š˜©š˜°š˜ø you ask, not just š˜øš˜©š˜° you ask. // The Agent Loop is Everything It's not just LLM ā†’ Output. Real agents iterate through reasoning ā†’ tool calls ā†’ observation ā†’ refinement. This dynamic execution path is what separates true agents from simple API wrappers. // Safety Guardrails Aren't Optional : They're Essential Read-only access, query validation, row limits, logs and error handling aren't afterthoughts. They're foundational. One misconfigured query can bring down your entire database. Period. // Memory Management Changes the Game Adding conversation history transforms single-shot queries into meaningful dialogue. Users can ask follow-ups naturally. But manage it wisely, unbounded memory explodes your token costs. // Evaluation & Fallback Strategies = Smart Cost Optimization Test your models systematically. Sometimes GPT-4 mini outperforms GPT-4 Turbo at 1/4 the cost. Build intelligent fallback mechanisms; simple models for routine queries, powerful ones for edge cases. A heartfelt thank you to Aishwarya Srinivasan, Arvind Narayanamurthy for setting up context/de-briefing about AI agent, and bringing in Rami Krispin (SME) for this Lightening session. Your generosity in sharing examples/use cases, live demos are highly appreciated. I gained additional insights and I'm sure all the participants who attended "live" would have benefitted, and reach out to discord community for Q&A/follow-up. #SQLAIAgents #AIEngineering #LLMs #DataAccess #ProductionAI #TechEducation #LinkedInLearning
From Text to SQL: How SQL AI Agents are Transforming Data Access Wrapped up an incredible session on š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ with Rami Krispin, where he shared his rich experience and production ready play-book. Encapsulated - Top 5 pointers (provided AI generated image for illustration only) //Context is King, Not the Model You don't need the fanciest LLM to build powerful SQL agents. A thoughtfully engineered prompt with rich context (schema details, categorical values, business logic) can outperform expensive models. It's about š˜©š˜°š˜ø you ask, not just š˜øš˜©š˜° you ask. // The Agent Loop is Everything It's not just LLM ā†’ Output. Real agents iterate through reasoning ā†’ tool calls ā†’ observation ā†’ refinement. This dynamic execution path is what separates true agents from simple API wrappers. // Safety Guardrails Aren't Optional : They're Essential Read-only access, query validation, row limits, logs and error handling aren't afterthoughts. They're foundational. One misconfigured query can bring down your entire database. Period. // Memory Management Changes the Game Adding conversation history transforms single-shot queries into meaningful dialogue. Users can ask follow-ups naturally. But manage it wisely, unbounded memory explodes your token costs. // Evaluation & Fallback Strategies = Smart Cost Optimization Test your models systematically. Sometimes GPT-4 mini outperforms GPT-4 Turbo at 1/4 the cost. Build intelligent fallback mechanisms; simple models for routine queries, powerful ones for edge cases. A heartfelt thank you to Aishwarya Srinivasan, Arvind Narayanamurthy for setting up context/de-briefing about AI agent, and bringing in Rami Krispin (SME) for this Lightening session. Your generosity in sharing examples/use cases, live demos are highly appreciated. I gained additional insights and I'm sure all the participants who attended "live" would have benefitted, and reach out to discord community for Q&A/follow-up. #SQLAIAgents #AIEngineering #LLMs #DataAccess #ProductionAI #TechEducation #LinkedInLearning

Yash Jaiswal

Power BI Developer and BI Specialist | 5 Years Enterprise BI | DAX | SQL | AI | Microsoft Fabric | Open to Senior Roles

Joined a free session today on Building SQL AI Agents and one thing was clear from the start. Getting an LLM to write SQL is easy. Getting it production ready is a completely different game. What actually separates a demo from a real system: autonomous context generation, SQL validation, safety checks , error handling and a debugger agent running in the background. The prompt isn't something you write once either. In production it gets assembled dynamically from your database schema, domain skills, and user question every single time. Shoutout to Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin for keeping it brutally honest and technically sharp.
Joined a free session today on Building SQL AI Agents and one thing was clear from the start. Getting an LLM to write SQL is easy. Getting it production ready is a completely different game. What actually separates a demo from a real system: autonomous context generation, SQL validation, safety checks , error handling and a debugger agent running in the background. The prompt isn't something you write once either. In production it gets assembled dynamically from your database schema, domain skills, and user question every single time. Shoutout to Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin for keeping it brutally honest and technically sharp.

Haadvitha Nimmagadda

Analytics Engineer | SQL | Power BI | Azure | Databricks

Attended a session on Building SQL AI Agents at The Gen Academy today. Honest take? Parts of it felt overwhelming. As someone just starting to explore AI, I could follow the high-level ideas, but the deeper discussions around context, architecture, and guardrails made one thing clear: Thereā€™s a lot to learn here. And that was the most useful part. Before writing a single line of code, the session broke down how agents actually work: ā€¢ Agents donā€™t respond in one step, they run a loop: Think ā†’ Act ā†’ Observe until the goal is met ā€¢ There are key levers you can control to shape that loop ā€¢ Every agent is built from a set of core building blocks ā€¢ Not every problem needs an agent, knowing when to use one matters Understanding these fundamentals before building matters more than I expected. Then it got more practical, what it takes to move from idea to something usable: ā€¢ The gap between a polished demo and a production system is significant ā€¢ Context is more than a prompt, it includes instructions, knowledge, and state ā€¢ Guardrails and validation are what make these systems reliable Instead of trying to learn everything upfront, Iā€™ve decided to start building a simple SQL agent and learn through the process. Thanks to Aishwarya Srinivasan, Arvind Narayanamurthy and Rami Krispin for keeping it practical and grounded. #AIAgents #GenAI #SQLAgent #DataEngineering #DataAnalytics
Attended a session on Building SQL AI Agents at The Gen Academy today. Honest take? Parts of it felt overwhelming. As someone just starting to explore AI, I could follow the high-level ideas, but the deeper discussions around context, architecture, and guardrails made one thing clear: Thereā€™s a lot to learn here. And that was the most useful part. Before writing a single line of code, the session broke down how agents actually work: ā€¢ Agents donā€™t respond in one step, they run a loop: Think ā†’ Act ā†’ Observe until the goal is met ā€¢ There are key levers you can control to shape that loop ā€¢ Every agent is built from a set of core building blocks ā€¢ Not every problem needs an agent, knowing when to use one matters Understanding these fundamentals before building matters more than I expected. Then it got more practical, what it takes to move from idea to something usable: ā€¢ The gap between a polished demo and a production system is significant ā€¢ Context is more than a prompt, it includes instructions, knowledge, and state ā€¢ Guardrails and validation are what make these systems reliable Instead of trying to learn everything upfront, Iā€™ve decided to start building a simple SQL agent and learn through the process. Thanks to Aishwarya Srinivasan, Arvind Narayanamurthy and Rami Krispin for keeping it practical and grounded. #AIAgents #GenAI #SQLAgent #DataEngineering #DataAnalytics

Vaishnavi Patil

Data Science & Analytics @ Exelon (BGE) | SQL ā€¢ Python ā€¢ ML ā€¢ Power BI/Fabric ā€¢ Tableau | Forecasting, KPI Insights, Databricks, Azure | Turning Data into Scalable Business Impact

šŸš€ Spent my weekend diving into Building SQL AI Agents ā€” and it was genuinely one of the most practical AI sessions Iā€™ve attended. Really enjoyed learning how LLMs + agents are moving from demos to real production systems ā€” especially: Natural language ā†’ SQL in real-world use cases Context engineering for reliable outputs Agent workflows (tools, planning, memory) Importance of guardrails & data access control Loved how everything tied back to actual data + enterprise systems, not just theory. Thank you Aishwarya Srinivasan Rami Krispin and Arvind Narayanamurthy! Looking forward to diving deeper into this space and building something hands-on soon. #AI #LLM #SQL #AIAgents #DataEngineering #RAG #TechLearning
šŸš€ Spent my weekend diving into Building SQL AI Agents ā€” and it was genuinely one of the most practical AI sessions Iā€™ve attended. Really enjoyed learning how LLMs + agents are moving from demos to real production systems ā€” especially: Natural language ā†’ SQL in real-world use cases Context engineering for reliable outputs Agent workflows (tools, planning, memory) Importance of guardrails & data access control Loved how everything tied back to actual data + enterprise systems, not just theory. Thank you Aishwarya Srinivasan Rami Krispin and Arvind Narayanamurthy! Looking forward to diving deeper into this space and building something hands-on soon. #AI #LLM #SQL #AIAgents #DataEngineering #RAG #TechLearning

Yuvansh Bhardwaj

AI Engineer | Data Scientist | Gen AI Specialist

Attended an incredible session on š—•š˜‚š—¶š—¹š—±š—¶š—»š—“ š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ hosted by Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin. š—œš—³ š—®š—“š—²š—»š˜š˜€ š—®š—暝—² š˜š—µš—² š—Æš—暝—®š—¶š—» š—¼š—³ š˜š—µš—² š—”š—œ š—²š—°š—¼š˜€š˜†š˜€š˜š—²š—ŗ, š˜š—µš—²š—» š—±š—®š˜š—® š—®š—»š—± š—®š—»š—®š—¹š˜†š˜š—¶š—°š˜€ š—®š—暝—² š˜š—µš—² š—³š˜‚š—²š—¹. So I went ahead and built my own SQL AI Agent from scratch. What I built: āœ… Natural language to SQL using Ollama (free, no API key) āœ… SQLite database with zero setup āœ… Streamlit UI with query history and CSV export āœ… Runs 100% locally, no cloud, no cost Huge thanks to Aishwarya, Arvind, and Rami for making this so hands-on. Sessions like this push you to stop watching and start building. Check out the repo here: šŸ”— lnkd.in/ebX4Dqix #AI #SQLAgent #Ollama #LLM #DataScience #ArtificialIntelligence #DataEngineering #MachineLearning #Python #Canada
Attended an incredible session on š—•š˜‚š—¶š—¹š—±š—¶š—»š—“ š—¦š—¤š—Ÿ š—”š—œ š—”š—“š—²š—»š˜š˜€ hosted by Aishwarya Srinivasan, Arvind Narayanamurthy, and Rami Krispin. š—œš—³ š—®š—“š—²š—»š˜š˜€ š—®š—暝—² š˜š—µš—² š—Æš—暝—®š—¶š—» š—¼š—³ š˜š—µš—² š—”š—œ š—²š—°š—¼š˜€š˜†š˜€š˜š—²š—ŗ, š˜š—µš—²š—» š—±š—®š˜š—® š—®š—»š—± š—®š—»š—®š—¹š˜†š˜š—¶š—°š˜€ š—®š—暝—² š˜š—µš—² š—³š˜‚š—²š—¹. So I went ahead and built my own SQL AI Agent from scratch. What I built: āœ… Natural language to SQL using Ollama (free, no API key) āœ… SQLite database with zero setup āœ… Streamlit UI with query history and CSV export āœ… Runs 100% locally, no cloud, no cost Huge thanks to Aishwarya, Arvind, and Rami for making this so hands-on. Sessions like this push you to stop watching and start building. Check out the repo here: šŸ”— lnkd.in/ebX4Dqix #AI #SQLAgent #Ollama #LLM #DataScience #ArtificialIntelligence #DataEngineering #MachineLearning #Python #Canada

Karthic Kumar Eswari Subramanian

Principal Engineer |AI/ML & AR/VR Explorer | Cloud & Microservices Expert|Full Stack Developer | DevOps Enthusiast | Tech Mentor |

Wow, just finished the "Vibe Coding: Code Smarter with AI" course by Aishwarya and Arvind, and I'm absolutely buzzing! This wasn't just another coding tutorial; it was a truly vibrant and insightful experience. The way Aishwarya and Arvind broke down how to leverage AI tools like Replit AI and Cursor AI to build practical applications was fantastic. Seeing a sample portfolio app come to life with Replit AI and then diving into a financial portfolio web app using Cursor AI really solidified the concepts. It wasn't just theoretical; it was hands-on and immediately applicable. What truly set this course apart was the engaging and patient way Aishwarya and Arvind addressed everyone's questions. No matter how big or small, they took the time to provide clear and helpful answers. Even when a few of us were clearly in "vibe nap" mode between coding sessions (we've all been there!), Aishwarya and Arvind had a knack for bringing us back into the flow and ensuring we didn't miss crucial points. It speaks volumes about their teaching style that even in those moments, the explanations were clear enough to still grasp the core ideas and improve our skills. The course content was incredibly well-structured, and the pace felt just right. You could tell that Aishwarya and Arvind are not only experts in their field but also passionate about sharing their knowledge. Their enthusiasm was infectious! And you know what? Forget complex algorithms and years of cryptic syntax! It turns out, channeling our inner Captain America, the secret weapon to building amazing applications with AI assistance boils down to one simple truth: "Whatever It Takes!" ...which, in this context, hilariously seems to mean just knowing plain English to tell the AI what you want. Who knew building the future could be this...linguistically straightforward? Thanks, Aishwarya and Arvind, for showing us the accessible path to code smarter! Aishwarya Srinivasan Arvind Narayanamurthy Deploying code failed and here is the pdf format of the vibe course portfolio that I built with the AI tool :)
Wow, just finished the "Vibe Coding: Code Smarter with AI" course by Aishwarya and Arvind, and I'm absolutely buzzing! This wasn't just another coding tutorial; it was a truly vibrant and insightful experience. The way Aishwarya and Arvind broke down how to leverage AI tools like Replit AI and Cursor AI to build practical applications was fantastic. Seeing a sample portfolio app come to life with Replit AI and then diving into a financial portfolio web app using Cursor AI really solidified the concepts. It wasn't just theoretical; it was hands-on and immediately applicable. What truly set this course apart was the engaging and patient way Aishwarya and Arvind addressed everyone's questions. No matter how big or small, they took the time to provide clear and helpful answers. Even when a few of us were clearly in "vibe nap" mode between coding sessions (we've all been there!), Aishwarya and Arvind had a knack for bringing us back into the flow and ensuring we didn't miss crucial points. It speaks volumes about their teaching style that even in those moments, the explanations were clear enough to still grasp the core ideas and improve our skills. The course content was incredibly well-structured, and the pace felt just right. You could tell that Aishwarya and Arvind are not only experts in their field but also passionate about sharing their knowledge. Their enthusiasm was infectious! And you know what? Forget complex algorithms and years of cryptic syntax! It turns out, channeling our inner Captain America, the secret weapon to building amazing applications with AI assistance boils down to one simple truth: "Whatever It Takes!" ...which, in this context, hilariously seems to mean just knowing plain English to tell the AI what you want. Who knew building the future could be this...linguistically straightforward? Thanks, Aishwarya and Arvind, for showing us the accessible path to code smarter! Aishwarya Srinivasan Arvind Narayanamurthy Deploying code failed and here is the pdf format of the vibe course portfolio that I built with the AI tool :)

Balavinayagam Shanmugavel

Architecture & Engineering | Core Banking Ā· Lending Ā· Private & Consumer Banking Ā· Loyalty | Microservices Ā· AWS Ā· Event-Driven Architecture | Generative & Agentic AI

RAG Workshop Takeaways ā€“ Powered by Aishwarya Srinivasan & Arvind Narayanamurthy Recently, I had the opportunity to attend an insightful online RAG workshop conducted by Aishwarya Srinivasan and Arvind Narayanamurthy on July 26 (11 PM) to July 27 (2 AM) SGT. Despite the late hours, it was absolutely worth staying up for ā€” the session was packed with conceptual, practical knowledge and real-world techniques! Here are my key takeaways: šŸ”¹ RAG (Retrieval-Augmented Generation) Fundamentals ā†’ Deep dive into how RAG augments LLMs by injecting fresh, contextual data at runtime, overcoming static training limitations. šŸ”¹ Live Demo ā†’ A complete end-to-end RAG pipeline demo using LangChain and Pinecone ā€“ from ingestion to generation. šŸ”¹ Hands-On: Building a Basic RAG Workflow ā†’ Pipeline Steps: Data Extraction ā€“ From a live website Chunking ā€“ Breaking down into manageable segments Indexing ā€“ Using ChromaDB and Google's Generative AI Embeddings Retrieval ā€“ Based on vector similarity Generation ā€“ Powered by Gemini 1.5 Flash šŸ”¹ Metadata-Aware RAG ā†’ Enhancing retrieval precision by extracting and indexing metadata alongside content for filtered, targeted queries. šŸ”¹ Hybrid RAG Architecture ā†’ Combining semantic search (vector-based) with lexical/keyword search for balanced relevance and coverage. šŸ”¹ Similarity Techniques ā†’ Leveraging cosine similarity to improve document-query matching effectiveness. šŸ”¹ Architecture Patterns & Debugging Tips ā†’ Field-tested practices and design examples for building production-grade RAG systems. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for this power packed session, looking forward for more hands-on sessions šŸ‘ #RAG #AI #LLM
RAG Workshop Takeaways ā€“ Powered by Aishwarya Srinivasan & Arvind Narayanamurthy Recently, I had the opportunity to attend an insightful online RAG workshop conducted by Aishwarya Srinivasan and Arvind Narayanamurthy on July 26 (11 PM) to July 27 (2 AM) SGT. Despite the late hours, it was absolutely worth staying up for ā€” the session was packed with conceptual, practical knowledge and real-world techniques! Here are my key takeaways: šŸ”¹ RAG (Retrieval-Augmented Generation) Fundamentals ā†’ Deep dive into how RAG augments LLMs by injecting fresh, contextual data at runtime, overcoming static training limitations. šŸ”¹ Live Demo ā†’ A complete end-to-end RAG pipeline demo using LangChain and Pinecone ā€“ from ingestion to generation. šŸ”¹ Hands-On: Building a Basic RAG Workflow ā†’ Pipeline Steps: Data Extraction ā€“ From a live website Chunking ā€“ Breaking down into manageable segments Indexing ā€“ Using ChromaDB and Google's Generative AI Embeddings Retrieval ā€“ Based on vector similarity Generation ā€“ Powered by Gemini 1.5 Flash šŸ”¹ Metadata-Aware RAG ā†’ Enhancing retrieval precision by extracting and indexing metadata alongside content for filtered, targeted queries. šŸ”¹ Hybrid RAG Architecture ā†’ Combining semantic search (vector-based) with lexical/keyword search for balanced relevance and coverage. šŸ”¹ Similarity Techniques ā†’ Leveraging cosine similarity to improve document-query matching effectiveness. šŸ”¹ Architecture Patterns & Debugging Tips ā†’ Field-tested practices and design examples for building production-grade RAG systems. Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for this power packed session, looking forward for more hands-on sessions šŸ‘ #RAG #AI #LLM

Sunny Rodrigues

āœšŸ» student forever // onboarding + project ops + customer success + product @ scale (saas, ai, social) // google + meta alum // vibe-coder // built on making a difference & curiosity in šŸŒ

I spent my Saturday at the Context Engineering Workshop with The Gen Academy, and hereā€™s the thing. Some sessions simply teach you something. This one actually shifted how I think. It was all about Prompt Engineering until I heard about Context Engineering. So what is it ?? Context engineering sits at the heart of building reliable AI systems, but very few people break it down in a way thatā€™s practical, structured, and immediately usable. This workshop did exactly that. Huge thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and the entire Gen Academy team for creating a space where learning feels both rigorous and accessible. I walked away with frameworks I can apply right away, along with a sharper mental model for how to design context in a way that elevates performance, clarity, and trust. For anyone exploring agentic workflows, retrieval, or prompt engineering at a deeper level, the teaching and the materials are an amazing source for both technical and non-technical folks. Grateful to have been part of this one and excited to keep sharpening this skillset. #ai #contextengineering #machinelearning #agents #llms #genai #community #learning #sanfrancisco #sftech
I spent my Saturday at the Context Engineering Workshop with The Gen Academy, and hereā€™s the thing. Some sessions simply teach you something. This one actually shifted how I think. It was all about Prompt Engineering until I heard about Context Engineering. So what is it ?? Context engineering sits at the heart of building reliable AI systems, but very few people break it down in a way thatā€™s practical, structured, and immediately usable. This workshop did exactly that. Huge thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and the entire Gen Academy team for creating a space where learning feels both rigorous and accessible. I walked away with frameworks I can apply right away, along with a sharper mental model for how to design context in a way that elevates performance, clarity, and trust. For anyone exploring agentic workflows, retrieval, or prompt engineering at a deeper level, the teaching and the materials are an amazing source for both technical and non-technical folks. Grateful to have been part of this one and excited to keep sharpening this skillset. #ai #contextengineering #machinelearning #agents #llms #genai #community #learning #sanfrancisco #sftech

LI
Lavanya Nallamshetty, PMP, ITIL

Sr. Infrastructure Project Manager | Founder of LucidSpec | Cloud | Data Center | Zero Trust | AIOps

Attended an incredible RAG Workshop over the weekend! The trainers, Aishwarya Srinivasan and Arvind Narayanamurthy did a fantastic job explaining the concepts clearly and breaking down complex ideas in such an accessible way. The workshop was thoughtfully structured, with a perfect balance of theory and hands-on practice, which made the learning experience so much more effective. If youā€™re interested in Retrieval-Augmented Generation (RAG) and building real world AI applications, this is something to checkout. Part 2 and Part 3 are coming up soon I believe.
Attended an incredible RAG Workshop over the weekend! The trainers, Aishwarya Srinivasan and Arvind Narayanamurthy did a fantastic job explaining the concepts clearly and breaking down complex ideas in such an accessible way. The workshop was thoughtfully structured, with a perfect balance of theory and hands-on practice, which made the learning experience so much more effective. If youā€™re interested in Retrieval-Augmented Generation (RAG) and building real world AI applications, this is something to checkout. Part 2 and Part 3 are coming up soon I believe.

Geetha priya J

AI Engineer @American Express

I attended a hands-on workshop over the weekend on LangChain and LangGraph led by Aishwarya Srinivasan and Arvind Narayanamurthy, and it was an incredibly valuable learning experience. The session was well structured, starting with foundational concepts and gradually moving into practical, real-world implementations of LangChain and LangGraph, clearly explaining how and when to use each. LangGraph is a framework for building stateful, controllable, graph-based AI workflows.  ā€¢ It lets you design LLM applications as graphs with nodes, edges, and shared state, enabling:  ā€¢ Memory across steps for long, multi-step workflows  ā€¢ Precise control over how an agent thinks and acts  ā€¢ Safe, structured tool calls inside workflows LangChain, on the other hand, focuses on building production-ready LLM applications by providing:  ā€¢ Components for prompts, models, chains, tools  ā€¢ Simplified orchestration of LLM logic and memory  ā€¢ Standardized patterns for real-world use cases A practical, hands-on session that seamlessly connected concepts with implementation. Huge thanks to Aishwarya and Arvind for delivering such an engaging and well-executed workshop. #LangChain #LangGraph #LLM #AIEngineering #AgenticAI #Learning #AI #workshop
I attended a hands-on workshop over the weekend on LangChain and LangGraph led by Aishwarya Srinivasan and Arvind Narayanamurthy, and it was an incredibly valuable learning experience. The session was well structured, starting with foundational concepts and gradually moving into practical, real-world implementations of LangChain and LangGraph, clearly explaining how and when to use each. LangGraph is a framework for building stateful, controllable, graph-based AI workflows.  ā€¢ It lets you design LLM applications as graphs with nodes, edges, and shared state, enabling:  ā€¢ Memory across steps for long, multi-step workflows  ā€¢ Precise control over how an agent thinks and acts  ā€¢ Safe, structured tool calls inside workflows LangChain, on the other hand, focuses on building production-ready LLM applications by providing:  ā€¢ Components for prompts, models, chains, tools  ā€¢ Simplified orchestration of LLM logic and memory  ā€¢ Standardized patterns for real-world use cases A practical, hands-on session that seamlessly connected concepts with implementation. Huge thanks to Aishwarya and Arvind for delivering such an engaging and well-executed workshop. #LangChain #LangGraph #LLM #AIEngineering #AgenticAI #Learning #AI #workshop

Laura Ruiz

Associate Director, Data Science @AstraZeneca

Just a few days ago, I spent nearly 5 hours in an incredible hands-on workshop on š—”š—“š—²š—»š˜š—¶š—° š—”š—œ š—Ŗš—¼š—暝—øš—³š—¹š—¼š˜„š˜€ š—®š—»š—± š— š—¼š—±š—²š—¹ š—–š—¼š—»š˜š—²š˜…š˜ š—£š—暝—¼š˜š—¼š—°š—¼š—¹ (š— š—–š—£) implementation! Led by the brilliant Aishwarya Srinivasan and Arvind Narayanamurthy, this session was honestly one of the clearest explanations of this topic I've ever seen so far. We dove deep into: - š— š—–š—£: What Model Context Protocol (MCP) is, distinguishing its role in standardizing AI agent interactions from traditional function calling. We also covered the nuances of its transport layers: stdio (for co-located client-server) vs. sse (for pre-hosted server connections). - š—Ÿš—®š—»š—“š—šš—暝—®š—½š—µ š˜ƒš˜€. š—Ÿš—®š—»š—“š—–š—µš—®š—¶š—»: Explored their distinct advantages, particularly how LangGraph facilitates building more stateful, non-linear workflows using Agentic patterns like ReAct (Reason + Act) and Reflection, which is crucial for complex, adaptive agents. - š—„š—”š—š: Explanation and common misconceptions around Retrieval-Augmented Generation (RAG), emphasizing that it's distinct from large context windows and function calling. We saw how RAG ensures factual integrity and contextual grounding by integrating external knowledge stores. - š—”š—“š—²š—»š˜š—¶š—° š—£š—®š˜š˜š—²š—暝—»š˜€ & š—˜š˜…š—²š—°š˜‚š˜š—¶š—¼š—»: A deep dive into patterns like ReAct and Reflection, empowering agents to autonomously plan and execute multi-step actions. - š—£š—暝—®š—°š˜š—¶š—°š—®š—¹ š—œš—ŗš—½š—¹š—²š—ŗš—²š—»š˜š—®š˜š—¶š—¼š—»: Live coding sessions, including building an MCP server from scratch and a sample product support agent using Gemini models. - š—¦š—²š—°š˜‚š—暝—¶š˜š˜† & š—¢š—Æš˜€š—²š—暝˜ƒš—®š—Æš—¶š—¹š—¶š˜š˜†: Discussed critical aspects of agent safety, from preventing prompt injection to implementing human-in-the-loop approvals, and touched upon tools for LLM observability in production environments. If you have a technical background and want to deep dive on these concepts, I recommend checking out the workshop recording. It's available for a very reasonable price here: lnkd.in/dAU7VvCU Huge thanks again to Aishwarya Srinivasan and Arvind Narayanamurthy for such an impactful session! #AgenticAI #LangGraph #Langsmith #ZapierAI #MCP #RAG #ReActPattern #ReflectionPattern
Just a few days ago, I spent nearly 5 hours in an incredible hands-on workshop on š—”š—“š—²š—»š˜š—¶š—° š—”š—œ š—Ŗš—¼š—暝—øš—³š—¹š—¼š˜„š˜€ š—®š—»š—± š— š—¼š—±š—²š—¹ š—–š—¼š—»š˜š—²š˜…š˜ š—£š—暝—¼š˜š—¼š—°š—¼š—¹ (š— š—–š—£) implementation! Led by the brilliant Aishwarya Srinivasan and Arvind Narayanamurthy, this session was honestly one of the clearest explanations of this topic I've ever seen so far. We dove deep into: - š— š—–š—£: What Model Context Protocol (MCP) is, distinguishing its role in standardizing AI agent interactions from traditional function calling. We also covered the nuances of its transport layers: stdio (for co-located client-server) vs. sse (for pre-hosted server connections). - š—Ÿš—®š—»š—“š—šš—暝—®š—½š—µ š˜ƒš˜€. š—Ÿš—®š—»š—“š—–š—µš—®š—¶š—»: Explored their distinct advantages, particularly how LangGraph facilitates building more stateful, non-linear workflows using Agentic patterns like ReAct (Reason + Act) and Reflection, which is crucial for complex, adaptive agents. - š—„š—”š—š: Explanation and common misconceptions around Retrieval-Augmented Generation (RAG), emphasizing that it's distinct from large context windows and function calling. We saw how RAG ensures factual integrity and contextual grounding by integrating external knowledge stores. - š—”š—“š—²š—»š˜š—¶š—° š—£š—®š˜š˜š—²š—暝—»š˜€ & š—˜š˜…š—²š—°š˜‚š˜š—¶š—¼š—»: A deep dive into patterns like ReAct and Reflection, empowering agents to autonomously plan and execute multi-step actions. - š—£š—暝—®š—°š˜š—¶š—°š—®š—¹ š—œš—ŗš—½š—¹š—²š—ŗš—²š—»š˜š—®š˜š—¶š—¼š—»: Live coding sessions, including building an MCP server from scratch and a sample product support agent using Gemini models. - š—¦š—²š—°š˜‚š—暝—¶š˜š˜† & š—¢š—Æš˜€š—²š—暝˜ƒš—®š—Æš—¶š—¹š—¶š˜š˜†: Discussed critical aspects of agent safety, from preventing prompt injection to implementing human-in-the-loop approvals, and touched upon tools for LLM observability in production environments. If you have a technical background and want to deep dive on these concepts, I recommend checking out the workshop recording. It's available for a very reasonable price here: lnkd.in/dAU7VvCU Huge thanks again to Aishwarya Srinivasan and Arvind Narayanamurthy for such an impactful session! #AgenticAI #LangGraph #Langsmith #ZapierAI #MCP #RAG #ReActPattern #ReflectionPattern

Lakshmi Priya Gayatri Vutukuri

Data Scientist @ World Bank | Data Science | AI | ML | NLP | Azure

āœØ I attended an incredible hands-on workshop over the weekend on Agentic AI Workflows and Model Context Protocol (MCP) implementation led by Aishwarya Srinivasan and Arvind Narayanamurthy! It was such a valuable and enriching learning experience. This session was a deep dive into MCP and Agentic AI systems. We started from the basics and worked our way up to building an MCP server live, creating AI agents from scratch. šŸ¤–āš™ļø I took away so much from this session, but here are some of the key things I have learned: āœ… What MCP is and how it is different from function calling, enabling AI agent interactions. āœ… The transport layers in MCP - understanding when to use stdio (when the client spawns the server making them both in the same location) vs. sse (when the MCP server is already hosted and the client connects to it). šŸ”ŒšŸŒ āœ… Differences and use cases of LangGraph vs. LangChain, and how LangGraph allows building more stateful workflows using Agentic patterns like ReAct (Reason + Act) and Reflection. šŸ”„šŸ§  āœ… Understanding the trade-off between autonomy and control in AI systems and how to design for the right balance depending on the use case. āš–ļøšŸ¤” āœ… Building an MCP server from scratch šŸ› ļø āœ… Practical implementation of a real-world agent use case using LangGraph and ReAct Pattern šŸŒšŸ’” What stood out to me personally was how the session cleared up common misconceptions around RAG, function-calling, context windows, and how MCP fits into the bigger picture. The live coding part was a game-changer šŸ’»šŸš€ ā€” seeing things in action made the concepts click and has given me much more confidence in using these tools in the future. Huge thanks to Aishwarya and Arvind, for driving this fantastic initiative and delivering such an engaging and practical session šŸ™ŒšŸ‘. I truly enjoyed the energy, the depth of content, and the clarity you both brought to complex topics. Itā€™s always exciting to keep learning in this fast-evolving field šŸŒŸšŸ“š. #AgenticAI #LangGraph #LangChain #MCP #AIagents #ReActPattern #ReflectionPattern #AIworkflows #RAG #AutonomousAgents #AgenticPatterns
āœØ I attended an incredible hands-on workshop over the weekend on Agentic AI Workflows and Model Context Protocol (MCP) implementation led by Aishwarya Srinivasan and Arvind Narayanamurthy! It was such a valuable and enriching learning experience. This session was a deep dive into MCP and Agentic AI systems. We started from the basics and worked our way up to building an MCP server live, creating AI agents from scratch. šŸ¤–āš™ļø I took away so much from this session, but here are some of the key things I have learned: āœ… What MCP is and how it is different from function calling, enabling AI agent interactions. āœ… The transport layers in MCP - understanding when to use stdio (when the client spawns the server making them both in the same location) vs. sse (when the MCP server is already hosted and the client connects to it). šŸ”ŒšŸŒ āœ… Differences and use cases of LangGraph vs. LangChain, and how LangGraph allows building more stateful workflows using Agentic patterns like ReAct (Reason + Act) and Reflection. šŸ”„šŸ§  āœ… Understanding the trade-off between autonomy and control in AI systems and how to design for the right balance depending on the use case. āš–ļøšŸ¤” āœ… Building an MCP server from scratch šŸ› ļø āœ… Practical implementation of a real-world agent use case using LangGraph and ReAct Pattern šŸŒšŸ’” What stood out to me personally was how the session cleared up common misconceptions around RAG, function-calling, context windows, and how MCP fits into the bigger picture. The live coding part was a game-changer šŸ’»šŸš€ ā€” seeing things in action made the concepts click and has given me much more confidence in using these tools in the future. Huge thanks to Aishwarya and Arvind, for driving this fantastic initiative and delivering such an engaging and practical session šŸ™ŒšŸ‘. I truly enjoyed the energy, the depth of content, and the clarity you both brought to complex topics. Itā€™s always exciting to keep learning in this fast-evolving field šŸŒŸšŸ“š. #AgenticAI #LangGraph #LangChain #MCP #AIagents #ReActPattern #ReflectionPattern #AIworkflows #RAG #AutonomousAgents #AgenticPatterns

Rushabh Dev Kashyap

VP, Data Strategy for AI & Governance @ JPMC | ex-IBMer | Data platforms, SaaS/PaaS products, Big Data

šŸš€ Just attended a mind-blowing workshop on Agentic AI Workflows and Model Context Protocol (MCP) implementation with the incredible Aishwarya Srinivasan and Arvind Narayanamurthy. Truly one of the most insightful and energising sessions Iā€™ve been part of! It was a deep dive into the evolving world of #LLMs, #RAG, and #AgenticAI. I asked Perplexity to summarise some notes I took down during the workshop for generic consumption and here's what stood out: šŸ”¹ LangChain vs LangGraph LangChain offers a powerful abstraction layer to build AI apps quickly. LangGraph enables stateful workflows and is ideal for unstructured data and nested/non-linear logic ā€” perfect for building complex agents! šŸ”¹ RAG (Retrieval-Augmented Generation) decoded: Components: Query processing, Embedding, Vector Store, Context Integration, Response ā—Why RAG? To ensure factual integrity, contextual grounding, and for complete ownership on knowledge store. Misconception: A large context window ā‰  RAG. Nor is function calling the same as RAG! šŸ”¹ MCP (Model Context Protocol) in action: Standardizes communication between LLMs and external tools via a server-client architecture. ā—An MCP server can be connected to a single tool or package multiple tools together (e.g., Gmail, Slack, Google Drive). The client makes requests and knows which server/tool to use. ā—Transport layers: stdio (client-spawned, co-located) and sse (pre-hosted, accessible remotely). āœ… Crucial: Sanitize outputs, use trusted servers, enforce OAuth/API key-based access. šŸ”¹ Agentic Patterns & Execution We explored the ReAct (Reason + Act) pattern and the Reflection pattern. These patterns empower agents to autonomously plan and take multi-step actions. LangGraph makes these patterns stateful, solving a major limitation of stateless LLMs. šŸ”¹ Using Prebuilt Agents for building typical ReAct patterns - from langgraph.prebuilt import create_react_agent from langchain_google_genai import ChatGoogleGenerativeAI Built a sample Product support agent to with reasoning support using Gemini models. (gemini-2.5-flash-preview-04-17) šŸ”¹ Security & Observability Agent safety isnā€™t just a buzzword: From preventing prompt injection to human-in-the-loop approvals ā€” it's foundational. šŸ§  Wondering: What are some other tools like LangSmith that help with LLM observability in production environments? šŸ› ļø Langchain-MCP adapters enable tool compatibility across ecosystems. šŸ“Š Fine-tuning, prompt design, function calling ā€” each has a role in aligning model behavior to business needs. Last but not the least, thoroughly enjoyed the engaging conversations with fellow enthusiasts and learners from across the world! Cheers! #AI #MachineLearning #LLM #LangChain #LangGraph #MCP #AgenticAI #RAG #Gemini #PromptEngineering #FunctionCalling #AutonomousAgents #AIWorkflows #EnterpriseAI #VectorSearch #OpenAI #Anthropic #ZapierAI #Langsmith #Observability #SecureAI #DataAndAI
šŸš€ Just attended a mind-blowing workshop on Agentic AI Workflows and Model Context Protocol (MCP) implementation with the incredible Aishwarya Srinivasan and Arvind Narayanamurthy. Truly one of the most insightful and energising sessions Iā€™ve been part of! It was a deep dive into the evolving world of #LLMs, #RAG, and #AgenticAI. I asked Perplexity to summarise some notes I took down during the workshop for generic consumption and here's what stood out: šŸ”¹ LangChain vs LangGraph LangChain offers a powerful abstraction layer to build AI apps quickly. LangGraph enables stateful workflows and is ideal for unstructured data and nested/non-linear logic ā€” perfect for building complex agents! šŸ”¹ RAG (Retrieval-Augmented Generation) decoded: Components: Query processing, Embedding, Vector Store, Context Integration, Response ā—Why RAG? To ensure factual integrity, contextual grounding, and for complete ownership on knowledge store. Misconception: A large context window ā‰  RAG. Nor is function calling the same as RAG! šŸ”¹ MCP (Model Context Protocol) in action: Standardizes communication between LLMs and external tools via a server-client architecture. ā—An MCP server can be connected to a single tool or package multiple tools together (e.g., Gmail, Slack, Google Drive). The client makes requests and knows which server/tool to use. ā—Transport layers: stdio (client-spawned, co-located) and sse (pre-hosted, accessible remotely). āœ… Crucial: Sanitize outputs, use trusted servers, enforce OAuth/API key-based access. šŸ”¹ Agentic Patterns & Execution We explored the ReAct (Reason + Act) pattern and the Reflection pattern. These patterns empower agents to autonomously plan and take multi-step actions. LangGraph makes these patterns stateful, solving a major limitation of stateless LLMs. šŸ”¹ Using Prebuilt Agents for building typical ReAct patterns - from langgraph.prebuilt import create_react_agent from langchain_google_genai import ChatGoogleGenerativeAI Built a sample Product support agent to with reasoning support using Gemini models. (gemini-2.5-flash-preview-04-17) šŸ”¹ Security & Observability Agent safety isnā€™t just a buzzword: From preventing prompt injection to human-in-the-loop approvals ā€” it's foundational. šŸ§  Wondering: What are some other tools like LangSmith that help with LLM observability in production environments? šŸ› ļø Langchain-MCP adapters enable tool compatibility across ecosystems. šŸ“Š Fine-tuning, prompt design, function calling ā€” each has a role in aligning model behavior to business needs. Last but not the least, thoroughly enjoyed the engaging conversations with fellow enthusiasts and learners from across the world! Cheers! #AI #MachineLearning #LLM #LangChain #LangGraph #MCP #AgenticAI #RAG #Gemini #PromptEngineering #FunctionCalling #AutonomousAgents #AIWorkflows #EnterpriseAI #VectorSearch #OpenAI #Anthropic #ZapierAI #Langsmith #Observability #SecureAI #DataAndAI

SP
Sunil Pentapati

Technology Executive | Strategy | Data and AI Platform Engineering | Thought Leadership

Midnight coding hits differently, and itā€™s been a while. This Saturday, I felt that vibe again. :) I attended a Vibe coding workshop hosted by Aishwarya and Arvind in partnership with Replit. Using Replit and Cursor, I built a UI and spun up a Streamlit project. From my experience, scaling anything successfully requires: - A framework to streamline  - Patterns and best practices to drive reusability, efficiency, and adoption - Techniques to operationalize through automation and orchestration, enabling measurable and observable outcomes I came out with a few takeaways in each of these areas. #vibecoding #learning #streamlit
Midnight coding hits differently, and itā€™s been a while. This Saturday, I felt that vibe again. :) I attended a Vibe coding workshop hosted by Aishwarya and Arvind in partnership with Replit. Using Replit and Cursor, I built a UI and spun up a Streamlit project. From my experience, scaling anything successfully requires: - A framework to streamline  - Patterns and best practices to drive reusability, efficiency, and adoption - Techniques to operationalize through automation and orchestration, enabling measurable and observable outcomes I came out with a few takeaways in each of these areas. #vibecoding #learning #streamlit

Pawan Simha

Student @ SNPSU | Gemini Certified | Aspiring AI Product Manager | Python ā€¢ Machine Learning ā€¢ Web Development | CR & HRD Coordinator

AI isnā€™t the problem anymore. Context is. Yesterday night I attended the 1st Lightning Lesson on Retrieval-Augmented Generation (RAG) by Aishwarya Srinivasan and Arvind Narayanamurthy (Maven Ɨ The Gen Academy ). And it completely changed how I think about building AI systems. Instead of relying on static training data, RAG enables systems to retrieve relevant information in real time, producing responses grounded in actual, verifiable data. What stood out wasnā€™t just the concept, but the system design behind it: 1. How embeddings capture semantic meaning. 2. How chunking improves retrieval quality. 3. How vector databases power fast similarity search. It made one thing clear: The future of AI isnā€™t just about bigger models. Itā€™s about building systems that combine retrieval, context, and reasoning effectively. Lately, Iā€™ve been diving deeper into how these architectures work under the hood, and the more I explore, the more compelling this space becomes. Iā€™m currently exploring this further and planning to build something around it soon šŸš€ Question for my Network: For those working with LLMs, what vector databases are you using in production? #GenerativeAI #RAG #VectorDatabases #AIProductManagement #BuildInPublic #MCA #TechInnovation #SNPSU #BeyondAcademics
AI isnā€™t the problem anymore. Context is. Yesterday night I attended the 1st Lightning Lesson on Retrieval-Augmented Generation (RAG) by Aishwarya Srinivasan and Arvind Narayanamurthy (Maven Ɨ The Gen Academy ). And it completely changed how I think about building AI systems. Instead of relying on static training data, RAG enables systems to retrieve relevant information in real time, producing responses grounded in actual, verifiable data. What stood out wasnā€™t just the concept, but the system design behind it: 1. How embeddings capture semantic meaning. 2. How chunking improves retrieval quality. 3. How vector databases power fast similarity search. It made one thing clear: The future of AI isnā€™t just about bigger models. Itā€™s about building systems that combine retrieval, context, and reasoning effectively. Lately, Iā€™ve been diving deeper into how these architectures work under the hood, and the more I explore, the more compelling this space becomes. Iā€™m currently exploring this further and planning to build something around it soon šŸš€ Question for my Network: For those working with LLMs, what vector databases are you using in production? #GenerativeAI #RAG #VectorDatabases #AIProductManagement #BuildInPublic #MCA #TechInnovation #SNPSU #BeyondAcademics

Shivangi J.

GenAI Architect & Data Strategist | Building AI that thinks, acts, and serves people | Agentic AI ā€¢ RAG ā€¢ AI Safety āœØ | Y4D šŸ‡®šŸ‡³ šŸ‡ŗšŸ‡ø

šŸ™ Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for an outstanding workshop on RAG (Retrieval Augmented Generation). The depth of insights you both shared was incredible ā€” moving beyond the basics into real design patterns and enterprise-grade architectures: šŸ” Simple & Branched RAG for foundational retrieval šŸ¤” HyDE & Corrective RAG for reasoning and self-checks šŸ§  Agentic RAG with master/sub-agent orchestration & RBAC šŸŽØ Multimodal RAG handling text, images, audio, video at scale šŸ•ø Graph RAG enabling multi-hop reasoning & auditability through knowledge graphs I especially valued the deep dives into: šŸ§© Memory hierarchies (short-term, session, long-term) āš–ļø Compliance & traceability (audit trails, citations, observability tools) šŸ’° Cost trade-offs across multimodal patterns šŸ›  Tools in action ā€” Neo4j, LangChain, ChromaDB, OpenAI, OpenCLIP āœØ Key Takeaways: ā€¢ RAG is not just about chatbots ā€” itā€™s about building scalable, explainable, and compliant systems. ā€¢ Design patterns matter ā†’ Simple, Corrective, Agentic, Multimodal, and Graph RAG each unlock different capabilities. ā€¢ Memory hierarchies (short-term, session, long-term) are essential to balance cost, speed, and persistence. ā€¢ Compliance & auditability (via LangFuse, cloud-native tracing, or Graph provenance) must be built-in, not bolted on. ā€¢ Cost trade-offs are huge in multimodal RAG ā†’ the right design pattern can mean 10x efficiency. šŸ’” Questions Iā€™m still reflecting on: ā€¢ How do we best balance deterministic workflows with LLM-driven non-determinism in enterprise systems? ā€¢ In large-scale multimodal setups, how can we keep context fresh and relevant without overspending on tokens? ā€¢ Will Graph RAG (reasoning + provenance) or Agentic RAG (orchestration + sub-agents) dominate enterprise adoption next? Thank you both once again for expanding our thinking ā€” this workshop was a true masterclass. šŸ™Œ #RAG #GenAI #GraphRAG #MultimodalAI #AgenticAI #LLMs #AIEngineering
šŸ™ Huge thanks to Aishwarya Srinivasan and Arvind Narayanamurthy for an outstanding workshop on RAG (Retrieval Augmented Generation). The depth of insights you both shared was incredible ā€” moving beyond the basics into real design patterns and enterprise-grade architectures: šŸ” Simple & Branched RAG for foundational retrieval šŸ¤” HyDE & Corrective RAG for reasoning and self-checks šŸ§  Agentic RAG with master/sub-agent orchestration & RBAC šŸŽØ Multimodal RAG handling text, images, audio, video at scale šŸ•ø Graph RAG enabling multi-hop reasoning & auditability through knowledge graphs I especially valued the deep dives into: šŸ§© Memory hierarchies (short-term, session, long-term) āš–ļø Compliance & traceability (audit trails, citations, observability tools) šŸ’° Cost trade-offs across multimodal patterns šŸ›  Tools in action ā€” Neo4j, LangChain, ChromaDB, OpenAI, OpenCLIP āœØ Key Takeaways: ā€¢ RAG is not just about chatbots ā€” itā€™s about building scalable, explainable, and compliant systems. ā€¢ Design patterns matter ā†’ Simple, Corrective, Agentic, Multimodal, and Graph RAG each unlock different capabilities. ā€¢ Memory hierarchies (short-term, session, long-term) are essential to balance cost, speed, and persistence. ā€¢ Compliance & auditability (via LangFuse, cloud-native tracing, or Graph provenance) must be built-in, not bolted on. ā€¢ Cost trade-offs are huge in multimodal RAG ā†’ the right design pattern can mean 10x efficiency. šŸ’” Questions Iā€™m still reflecting on: ā€¢ How do we best balance deterministic workflows with LLM-driven non-determinism in enterprise systems? ā€¢ In large-scale multimodal setups, how can we keep context fresh and relevant without overspending on tokens? ā€¢ Will Graph RAG (reasoning + provenance) or Agentic RAG (orchestration + sub-agents) dominate enterprise adoption next? Thank you both once again for expanding our thinking ā€” this workshop was a true masterclass. šŸ™Œ #RAG #GenAI #GraphRAG #MultimodalAI #AgenticAI #LLMs #AIEngineering

Mohd Anas

Applied AI Engineer | Building RAG & Agentic AI Systems | LangChain ā€¢ LLMs ā€¢ GenAI | Open to Opportunities

Two weeks ago, I came across a post by Aishwarya Srinivasan One idea from that post stayed with me: donā€™t just learn AI ā€” build something real. It made me reflect on how I had been learning. Like many students, I was stuck in a loop of watching tutorials, taking courses, and saving posts, but not actually building anything meaningful. So I decided to change that. Instead of consuming more content, I picked a problem and started building from scratch. No perfect roadmap, no detailed plan ā€” just learning by doing and figuring things out along the way. Thatā€™s how I ended up building a Finance Agent powered by RAG (Retrieval-Augmented Generation): GitHub: lnkd.in/gz8_qJJP Live Demo: lnkd.in/gcQdPn7t The goal was to move beyond a simple chatbot and build something closer to a real AI system ā€” one that connects to actual financial data, retrieves relevant information, and generates grounded, contextual responses. While building this, I also explored something I hadnā€™t focused on before ā€” observability in AI systems. Using LangSmith, I started tracking how the system actually behaves: response accuracy, latency, and failure cases. Instead of just checking if the output ā€œlooks right,ā€ I could now measure how well the system was performing. I even ran multiple experiments comparing factual accuracy and latency, which helped me identify where the system was failing and improve it iteratively. This completely changed how I approached building. Instead of guessing, I could now iterate based on real data ā€” making the system not just functional, but more reliable. This experience made one thing very clear to me. Getting an AI model to work at a basic level is not that hard anymore. But building something that is reliable, useful, and closer to real-world applications ā€” thatā€™s where the real challenge (and learning) begins. The industry is clearly moving in this direction. Itā€™s no longer just about how powerful your model is, but about how well your system performs in real scenarios ā€” whether it can deliver value, scale, and be trusted. For me, this is just the beginning. If you're learning AI/ML right now, my biggest takeaway would be this: donā€™t wait to feel ready. Start building. Youā€™ll learn much faster by doing than by just consuming content. Would love to hear your thoughts or feedback on the project ā€” and curious to know, what are you building these days? #AI #GenAI #RAG #AgenticAI #MachineLearning #BuildInPublic #LearningInPublic
Two weeks ago, I came across a post by Aishwarya Srinivasan One idea from that post stayed with me: donā€™t just learn AI ā€” build something real. It made me reflect on how I had been learning. Like many students, I was stuck in a loop of watching tutorials, taking courses, and saving posts, but not actually building anything meaningful. So I decided to change that. Instead of consuming more content, I picked a problem and started building from scratch. No perfect roadmap, no detailed plan ā€” just learning by doing and figuring things out along the way. Thatā€™s how I ended up building a Finance Agent powered by RAG (Retrieval-Augmented Generation): GitHub: lnkd.in/gz8_qJJP Live Demo: lnkd.in/gcQdPn7t The goal was to move beyond a simple chatbot and build something closer to a real AI system ā€” one that connects to actual financial data, retrieves relevant information, and generates grounded, contextual responses. While building this, I also explored something I hadnā€™t focused on before ā€” observability in AI systems. Using LangSmith, I started tracking how the system actually behaves: response accuracy, latency, and failure cases. Instead of just checking if the output ā€œlooks right,ā€ I could now measure how well the system was performing. I even ran multiple experiments comparing factual accuracy and latency, which helped me identify where the system was failing and improve it iteratively. This completely changed how I approached building. Instead of guessing, I could now iterate based on real data ā€” making the system not just functional, but more reliable. This experience made one thing very clear to me. Getting an AI model to work at a basic level is not that hard anymore. But building something that is reliable, useful, and closer to real-world applications ā€” thatā€™s where the real challenge (and learning) begins. The industry is clearly moving in this direction. Itā€™s no longer just about how powerful your model is, but about how well your system performs in real scenarios ā€” whether it can deliver value, scale, and be trusted. For me, this is just the beginning. If you're learning AI/ML right now, my biggest takeaway would be this: donā€™t wait to feel ready. Start building. Youā€™ll learn much faster by doing than by just consuming content. Would love to hear your thoughts or feedback on the project ā€” and curious to know, what are you building these days? #AI #GenAI #RAG #AgenticAI #MachineLearning #BuildInPublic #LearningInPublic

Sai Harshitha Sivalingala

Software Engineer | Full-Stack & AI Automation | Built Open-Source tools used by 4.5K+ Devs | Python ā€¢ Java ā€¢ Node.js ā€¢ React ā€¢ AWS ā€¢ JavaScriptā€¢ Kafka

RAG worksā€¦ until you realize you have no idea what it's doing. šŸ§ āš ļø You can build a RAG pipeline, plug in embeddings, connect an LLM, and it works, but you have no idea what's actually happening inside. šŸ˜¶  Why is a query slow? What exactly got retrieved? Why did the model give a bad answer? Where do you even start improving? Everything just feels like a black box. While exploring this, I came across Verba (7.6k ā­ ), a really solid open-source RAG built on Weaviate. It already had real users, which made it the perfect system to build observability on top of. Github Link : lnkd.in/g7eNSVni I set up a reverse proxy in front of Verba and intercepted its WebSocket stream. That lets me track latency, capture retrieved chunks, cache responses, and log everything, all without modifying the actual RAG application. And this is where things got interesting. It's not just about measuring latency or errors. The metrics that actually matter are different: p90, p95 latencies (what users actually feel), slow queries (the ones that break trust), cache hit rate (cost + speed), and most importantly, thumbs down feedback. Because a system can be fast, have zero errors, and still give a completely wrong answer. Only the user tells you that. This made me realize, RAG isn't just about building pipelines. Itā€™s about making them observable, debuggable, and improvable. There's still one blind spot though the pre-retrieval phase (embeddings + search decisions). Thatā€™s what I'm planning to dig into next. Also, small reality check, I did use Claude to help initiallyā€¦ and it completely broke halfway šŸ˜….Ended up debugging a lot of weird edge cases myself, which turned out to be the most valuable part. Thanks to Aishwarya Srinivasan for planting the seed for this idea.šŸ™Œ šŸ”— GitHub: lnkd.in/gq8iscZ2 If you are working with RAG, you will probably hit this wall sooner or later.šŸ’­ #RAG #LLM #AIEngineering #AIObservability #OpenSource #SystemDesign #BuildInPublic #AIProjects
RAG worksā€¦ until you realize you have no idea what it's doing. šŸ§ āš ļø You can build a RAG pipeline, plug in embeddings, connect an LLM, and it works, but you have no idea what's actually happening inside. šŸ˜¶  Why is a query slow? What exactly got retrieved? Why did the model give a bad answer? Where do you even start improving? Everything just feels like a black box. While exploring this, I came across Verba (7.6k ā­ ), a really solid open-source RAG built on Weaviate. It already had real users, which made it the perfect system to build observability on top of. Github Link : lnkd.in/g7eNSVni I set up a reverse proxy in front of Verba and intercepted its WebSocket stream. That lets me track latency, capture retrieved chunks, cache responses, and log everything, all without modifying the actual RAG application. And this is where things got interesting. It's not just about measuring latency or errors. The metrics that actually matter are different: p90, p95 latencies (what users actually feel), slow queries (the ones that break trust), cache hit rate (cost + speed), and most importantly, thumbs down feedback. Because a system can be fast, have zero errors, and still give a completely wrong answer. Only the user tells you that. This made me realize, RAG isn't just about building pipelines. Itā€™s about making them observable, debuggable, and improvable. There's still one blind spot though the pre-retrieval phase (embeddings + search decisions). Thatā€™s what I'm planning to dig into next. Also, small reality check, I did use Claude to help initiallyā€¦ and it completely broke halfway šŸ˜….Ended up debugging a lot of weird edge cases myself, which turned out to be the most valuable part. Thanks to Aishwarya Srinivasan for planting the seed for this idea.šŸ™Œ šŸ”— GitHub: lnkd.in/gq8iscZ2 If you are working with RAG, you will probably hit this wall sooner or later.šŸ’­ #RAG #LLM #AIEngineering #AIObservability #OpenSource #SystemDesign #BuildInPublic #AIProjects

Yugank Das

Building with ML, AI and Web Design

I was on my usual evening YouTube scavenging session today when I accidentally stumbled across a video by Aishwarya Srinivasan explaining the fundamentals on RAG. I had no idea what RAG even was before watching it. It always seemed like a far fetched concept that would never make sense to someone so young in the industry as me. Nevertheless, I began scouring the web to get a hold of the basics, and unexpectedly, the concept clicked. I recently got back into spamming Blitz games at Chess.com, so I figured building a chess chatbot was the perfect way to learn by doing. RAG stands for Retrieval-Augmented Generation. Instead of just asking an AI a question and hoping it knows the answer, you first retrieve relevant passages from your own documents and then ask the model to answer using those. As a result, you usually receive more grounded and foundationally correct answers from the model. I used three chess PDFs covering openings, middlegames, and endgames as the knowledge base for this, and built the full pipeline from scratch. As usual, Groq came to my rescue as the LLM for the project with the Llama-3.3-70B-Versatile model. ChromaDB was the perfect choice for vector databases for first-timers like me. And well, it turns out chunk size actually matters more than I thought it did. My chunks were too small and kept splitting context mid-thought, so I doubled the size to 1000 characters with 200-character overlap and retrieval got a lot cleaner. Not all embedding models are the same. Swapping from all-MiniLM-L6-v2 to all-mpnet-base-v2 gave me better semantic search and suddenly the chatbot could find stuff it was completely missing before. Querying only n=3 chunks was too stingy. Bumping it to n=10 gave the LLM way more context to work with. With 3 retrieved chunks, it acted as if I asked it about pickleball and not chess. Printing chunk sources and metadata during debugging was a game changer. The model's final answer alone tells you nothing about what actually went wrong in the retrieval step. Doing this sped up my debug game by a mile. It is a naive RAG system by design. I have not implemented any reranking, memory or agents for the sake of simplicity so that I could fully focus on understanding what was happening under the hood. This is pretty much the same reason I repudiated the suggestion of using Streamlit to build a UI and stuck to a CLI (Command Line Interface), which is easier to debug around. Honestly one of the most satisfying 5-hour builds I have done. If you are curious about RAG and do not know where to start, go watch Aishwarya's video. #RAG #LLM #GenerativeAI #MachineLearning #LearningInPublic #Chess #AIProjects #BuildInPublic #NLP #VectorDatabase
I was on my usual evening YouTube scavenging session today when I accidentally stumbled across a video by Aishwarya Srinivasan explaining the fundamentals on RAG. I had no idea what RAG even was before watching it. It always seemed like a far fetched concept that would never make sense to someone so young in the industry as me. Nevertheless, I began scouring the web to get a hold of the basics, and unexpectedly, the concept clicked. I recently got back into spamming Blitz games at Chess.com, so I figured building a chess chatbot was the perfect way to learn by doing. RAG stands for Retrieval-Augmented Generation. Instead of just asking an AI a question and hoping it knows the answer, you first retrieve relevant passages from your own documents and then ask the model to answer using those. As a result, you usually receive more grounded and foundationally correct answers from the model. I used three chess PDFs covering openings, middlegames, and endgames as the knowledge base for this, and built the full pipeline from scratch. As usual, Groq came to my rescue as the LLM for the project with the Llama-3.3-70B-Versatile model. ChromaDB was the perfect choice for vector databases for first-timers like me. And well, it turns out chunk size actually matters more than I thought it did. My chunks were too small and kept splitting context mid-thought, so I doubled the size to 1000 characters with 200-character overlap and retrieval got a lot cleaner. Not all embedding models are the same. Swapping from all-MiniLM-L6-v2 to all-mpnet-base-v2 gave me better semantic search and suddenly the chatbot could find stuff it was completely missing before. Querying only n=3 chunks was too stingy. Bumping it to n=10 gave the LLM way more context to work with. With 3 retrieved chunks, it acted as if I asked it about pickleball and not chess. Printing chunk sources and metadata during debugging was a game changer. The model's final answer alone tells you nothing about what actually went wrong in the retrieval step. Doing this sped up my debug game by a mile. It is a naive RAG system by design. I have not implemented any reranking, memory or agents for the sake of simplicity so that I could fully focus on understanding what was happening under the hood. This is pretty much the same reason I repudiated the suggestion of using Streamlit to build a UI and stuck to a CLI (Command Line Interface), which is easier to debug around. Honestly one of the most satisfying 5-hour builds I have done. If you are curious about RAG and do not know where to start, go watch Aishwarya's video. #RAG #LLM #GenerativeAI #MachineLearning #LearningInPublic #Chess #AIProjects #BuildInPublic #NLP #VectorDatabase

Balaji Kottana

Data Science and Engineer| Snowflake | Spark| Databricks| SQL| Data Governance | Data Modeling| Tableau

Just wrapped up an incredibly insightful session on **LLM Fine-Tuning** with the amazing **Aishwarya** and **Arvind** on Topmate ā€” and I'm still processing how much ground we covered! šŸ™Œ From LoRA & QLoRA to Supervised Fine-Tuning, RAG optimization, and hands-on tooling with Unsloth and Hugging Face ā€” it was a masterclass in making large language models actually work for real-world use cases. But here's the thing that hit me hardest during the session: Looking at the forward pass diagram ā€” inputs, weights, weighted sums, activation functions ā€” I was reminded that **the fundamentals never go out of style.** In the age of agents and LLMs, it's easy to forget that everything still traces back to classical ML intuition. The math hasn't changed. The thinking hasn't changed. Only the scale has. šŸ”‘ The engineers who truly thrive in the agentic era won't be the ones who only prompt ā€” they'll be the ones who understand *why* the model behaves the way it does. Thank you **Aishwarya Srinivasan** and **Arvind Narayanamurthy** for not just teaching tools, but for reinforcing that deep understanding. Sessions like these are rare. If you haven't explored fine-tuning yet, now is the time. šŸš€ #LLMs #FineTuning #MachineLearning #AI #DeepLearning #LoRA #Topmate #ContinuousLearning
Just wrapped up an incredibly insightful session on **LLM Fine-Tuning** with the amazing **Aishwarya** and **Arvind** on Topmate ā€” and I'm still processing how much ground we covered! šŸ™Œ From LoRA & QLoRA to Supervised Fine-Tuning, RAG optimization, and hands-on tooling with Unsloth and Hugging Face ā€” it was a masterclass in making large language models actually work for real-world use cases. But here's the thing that hit me hardest during the session: Looking at the forward pass diagram ā€” inputs, weights, weighted sums, activation functions ā€” I was reminded that **the fundamentals never go out of style.** In the age of agents and LLMs, it's easy to forget that everything still traces back to classical ML intuition. The math hasn't changed. The thinking hasn't changed. Only the scale has. šŸ”‘ The engineers who truly thrive in the agentic era won't be the ones who only prompt ā€” they'll be the ones who understand *why* the model behaves the way it does. Thank you **Aishwarya Srinivasan** and **Arvind Narayanamurthy** for not just teaching tools, but for reinforcing that deep understanding. Sessions like these are rare. If you haven't explored fine-tuning yet, now is the time. šŸš€ #LLMs #FineTuning #MachineLearning #AI #DeepLearning #LoRA #Topmate #ContinuousLearning

Karthick Nethaji Kaleeswaran

Operations Director, Product Consulting @ HCLTech | Building AI-powered Product Consulting Practice | Agentic AI in enterprise | Strategy Consultant | PMI EVOLVE Speaker

I demo AI features to prospects almost every day. The reactions are genuinely exciting > "this is exactly what we need", "can we go live next month?", "our board has been asking for this". Then they go live. And two weeks later the feedback starts trickling in. Friction here. Unexpected behavior there. "It worked differently in the demo". Nothing broke. But nothing quite held either. I thought this was just my org figuring things out. Then I read this from Aishwarya Srinivasan. "The era of vibe-coding has brought us incredible speed, but it has also introduced a dangerous precedent in production: Vibe-based Evaluation. Vibing your way to production is the fastest path to architectural debt and user churn". That hit uncomfortably close to home. The demo works because we control everything. Production doesn't give us that courtesy. The shift I made: treat demo-readiness and production-readiness as parallel tracks, not sequential phases. The demo gets you in the door. Production discipline is what keeps you there. Shipping fast still matters. But shipping trustworthy is the actual moat. #ProductManagement #AIPM #EnterpriseAI #AIStrategy
I demo AI features to prospects almost every day. The reactions are genuinely exciting > "this is exactly what we need", "can we go live next month?", "our board has been asking for this". Then they go live. And two weeks later the feedback starts trickling in. Friction here. Unexpected behavior there. "It worked differently in the demo". Nothing broke. But nothing quite held either. I thought this was just my org figuring things out. Then I read this from Aishwarya Srinivasan. "The era of vibe-coding has brought us incredible speed, but it has also introduced a dangerous precedent in production: Vibe-based Evaluation. Vibing your way to production is the fastest path to architectural debt and user churn". That hit uncomfortably close to home. The demo works because we control everything. Production doesn't give us that courtesy. The shift I made: treat demo-readiness and production-readiness as parallel tracks, not sequential phases. The demo gets you in the door. Production discipline is what keeps you there. Shipping fast still matters. But shipping trustworthy is the actual moat. #ProductManagement #AIPM #EnterpriseAI #AIStrategy

Zeba Karkhanawala

Making sense of AI - at work, in books & on the internet | Reading & Curating my AI Book Curriculum 2026| AI @ Dell | Writes blogs @ lifeofzeba.com | AWAAI Austin Chapter Co-Lead | IIT Jodhpur | UMD

LLMs have always had this aura of mystery around them. Last Saturday, I tried to demystify at least one piece of it. Attended a workshop on fine-tuning LLMs with The Gen Academy, presented by Aishwarya Srinivasan and Arvind Narayanamurthy. I've read about fine-tuning in Chip Huyen's AI Engineering book, but it's such a complex topic that it's genuinely tough to wrap my head around. The workshop helped!! One of the key things I remembered from this workshop was LoRA architecture - instead of retraining a model from scratch, you freeze the original weights and add small trainable layers on top. Same outcome, way less compute. We also did a hands-on supervised fine-tuning lab where we fine-tuned Llama 3.2 on a doctor appointment booking chatbot. The task was intent classification: given a user message, can the model correctly identify what the person wants to do - book, cancel, list appointments. Coming from an ML background, the concept felt familiar but looking at it through an LLM lens was a different exercise - you're not training for fixed, predictable inputs. You're training for anything a user might say. Still have a long way to go with this topic. But Saturday was a good few hours of making it a little less opaque. #AI #LLMs #FineTuning
LLMs have always had this aura of mystery around them. Last Saturday, I tried to demystify at least one piece of it. Attended a workshop on fine-tuning LLMs with The Gen Academy, presented by Aishwarya Srinivasan and Arvind Narayanamurthy. I've read about fine-tuning in Chip Huyen's AI Engineering book, but it's such a complex topic that it's genuinely tough to wrap my head around. The workshop helped!! One of the key things I remembered from this workshop was LoRA architecture - instead of retraining a model from scratch, you freeze the original weights and add small trainable layers on top. Same outcome, way less compute. We also did a hands-on supervised fine-tuning lab where we fine-tuned Llama 3.2 on a doctor appointment booking chatbot. The task was intent classification: given a user message, can the model correctly identify what the person wants to do - book, cancel, list appointments. Coming from an ML background, the concept felt familiar but looking at it through an LLM lens was a different exercise - you're not training for fixed, predictable inputs. You're training for anything a user might say. Still have a long way to go with this topic. But Saturday was a good few hours of making it a little less opaque. #AI #LLMs #FineTuning

Yash Mishra

Sr. Associate at NPCI | AWS Certified | Azure Certified |Docker & Kubernetes | CI/CD | KCNA Certified | KCSA Certified | Gen AI

šŸš€ Attended an insightful session on Retrieval-Augmented Generation (RAG) Had the opportunity to attend a session focused on RAG, one of the most powerful approaches in modern AI systems. It was great to understand how combining retrieval mechanisms with LLMs can significantly improve accuracy and reduce hallucinations. Looking forward to exploring and implementing RAG in real-world use cases, especially in DevOps and AI-driven systems. ## @Aishwarya Srinivasan
šŸš€ Attended an insightful session on Retrieval-Augmented Generation (RAG) Had the opportunity to attend a session focused on RAG, one of the most powerful approaches in modern AI systems. It was great to understand how combining retrieval mechanisms with LLMs can significantly improve accuracy and reduce hallucinations. Looking forward to exploring and implementing RAG in real-world use cases, especially in DevOps and AI-driven systems. ## @Aishwarya Srinivasan

Sanket Kulkarni

GCPx2 ā€¢ Red Hatx2 | AI native Cloud Automation Engineer @ Google Operations | Python | SQL | Google Workspace | Networking | Linux | DevOps

šŸ” What happens when your AI confidently gives you the WRONG answer? . . . That's exactly what I uncovered today at the RAG Fundamentals lightning session by Aishwarya and Arvind. And it's not an LLM (ChatGPT, Gemini, Claude) problem. It's a RAG problem because they're trained on internet data, not your organization's internal docs, policies, or real-time databases. So when you ask ChatGPT "Can I share a file with our logistics vendor?" - Without thinking as per the rules, it not just gives you an affirmation but also tells you how to do it, and it violates the actual company policy. Classic hallucination šŸ˜µ This is exactly the gap RAG fills. āœ… Here's what I walked away understanding: šŸ’” What RAG actually does - It connects your LLM to knowledge it was never trained on and makes it cite its sources. You get a "grounded answer". šŸ§± How it works (generic architecture) - Documents > Chunks > Embeddings > Vector DB & the importance of Metadata, how numbers capture meaning, chunking strategies, re-ranking, etc. āš” Some myths around RAG - like Bigger context windows replace RAG, RAG only works on PDFs, Fine-tuning is better - is it šŸ¤” šŸ› ļø The hands-on demo - Covering a live RAG chatbot demo built on n8n using just one PDF as the knowledge source. The bot retrieved chunks from Vector DB, passed them through embeddings and answered questions with full source traceability. Watching the pipeline actually execute in real-time made every concept clear šŸ”„ šŸ“ And yes, RAG isn't magic: ā€¢ It's not fully real-time (there's a lag) ā€¢ Access control (RBAC) needs to be built on top - when we go deeper in complex RAG pipelines ā€¢ Garbage in = garbage out - your docs need to be good That was a jam-packed learning session and I can tell that every minute was worth it, Aishwarya and Arvind made sure of that šŸ’Æ More to come as I continue this journey into Agentic AI šŸš€ #RAG #GenerativeAI #AgenticAI #LLM #AIEngineering #MachineLearning #TheGenAcademy #AI #VectorDB #LangChain
šŸ” What happens when your AI confidently gives you the WRONG answer? . . . That's exactly what I uncovered today at the RAG Fundamentals lightning session by Aishwarya and Arvind. And it's not an LLM (ChatGPT, Gemini, Claude) problem. It's a RAG problem because they're trained on internet data, not your organization's internal docs, policies, or real-time databases. So when you ask ChatGPT "Can I share a file with our logistics vendor?" - Without thinking as per the rules, it not just gives you an affirmation but also tells you how to do it, and it violates the actual company policy. Classic hallucination šŸ˜µ This is exactly the gap RAG fills. āœ… Here's what I walked away understanding: šŸ’” What RAG actually does - It connects your LLM to knowledge it was never trained on and makes it cite its sources. You get a "grounded answer". šŸ§± How it works (generic architecture) - Documents > Chunks > Embeddings > Vector DB & the importance of Metadata, how numbers capture meaning, chunking strategies, re-ranking, etc. āš” Some myths around RAG - like Bigger context windows replace RAG, RAG only works on PDFs, Fine-tuning is better - is it šŸ¤” šŸ› ļø The hands-on demo - Covering a live RAG chatbot demo built on n8n using just one PDF as the knowledge source. The bot retrieved chunks from Vector DB, passed them through embeddings and answered questions with full source traceability. Watching the pipeline actually execute in real-time made every concept clear šŸ”„ šŸ“ And yes, RAG isn't magic: ā€¢ It's not fully real-time (there's a lag) ā€¢ Access control (RBAC) needs to be built on top - when we go deeper in complex RAG pipelines ā€¢ Garbage in = garbage out - your docs need to be good That was a jam-packed learning session and I can tell that every minute was worth it, Aishwarya and Arvind made sure of that šŸ’Æ More to come as I continue this journey into Agentic AI šŸš€ #RAG #GenerativeAI #AgenticAI #LLM #AIEngineering #MachineLearning #TheGenAcademy #AI #VectorDB #LangChain

Anand Sridharan

Service Delivery Transformation | Leading Intelligent Ops | Data & AI

š—Ŗš—µš˜† š—”š—œ š—›š—®š—¹š—¹š˜‚š—°š—¶š—»š—®š˜š—²š˜€ š—®š—»š—± š—›š—¼š˜„ š—„š—”š—š š—™š—¼š—暝—°š—²š˜€ š—§š—暝˜‚š˜š—µ? Attended š—„š—”š—š š—˜š˜…š—½š—¹š—®š—¶š—»š—²š—±: š—§š—µš—² š—”š—暝—°š—µš—¶š˜š—²š—°š˜š˜‚š—暝—² š—•š—²š—µš—¶š—»š—± š—”š—“š—²š—»š˜š—¶š—° š—”š—œ š—¦š˜†š˜€š˜š—²š—ŗš˜€ lightening session hosted by Aishwarya and Arvind packed with practical use cases, architecture insights, and scalable design patterns for Agentic AI. š—§š—µš—² š—›š—®š—暝—± š—§š—暝˜‚š˜š—µ:,Your LLM will confidently sabotage you unless you anchor it reality. š—„š—²š—®š—¹-š—Ŗš—¼š—暝—¹š—± š—˜š˜…š—®š—ŗš—½š—¹š—²: Ask "š˜Šš˜¢š˜Æ š˜ š˜“š˜©š˜¢š˜³š˜¦ š˜·š˜¦š˜Æš˜„š˜°š˜³ š˜§š˜Ŗš˜­š˜¦š˜“?"ā†’ ChatGPT says "š˜ š˜¦š˜“, š˜©š˜¦š˜³š˜¦'š˜“ š˜©š˜°š˜ø" Your Policy: "š˜•š˜° š˜¦š˜¹š˜µš˜¦š˜³š˜Æš˜¢š˜­ š˜“š˜©š˜¢š˜³š˜Ŗš˜Æš˜Ø š˜øš˜Ŗš˜µš˜©š˜°š˜¶š˜µ š˜¢š˜±š˜±š˜³š˜°š˜·š˜¢š˜­" The Gap: Internet-trained LLMs don't know your rules. š˜Žš˜¦š˜®š˜Ŗš˜Æš˜Ŗ-generated session summary image from my notes along with Key Takeaways: š— š˜†š˜š—µ š—•š˜‚š˜€š˜š—²š—±: Bigger context windows don't replace retrieval š—Ÿš—¶š˜ƒš—² š—£š—暝—¼š—¼š—³: n8n demo turned a PDF into a cited chatbot instantly š—„š—²š—®š—¹š—¶š˜š˜† š—–š—µš—²š—°š—ø: Lag exists, RBAC needs work, data quality rules all š—•š—¼š˜š˜š—¼š—ŗ š—Ÿš—¶š—»š—²: Stop playing with consumer AI. Build retrieval pipelines that cite your truth or bet your business on fabricated answers. Big thanks to Aishwarya and Arvind for an highly informative session with live demo and Q&A. P.S: Personal views. #EnterpriseAI #RAG #AgenticAI #AIGovernance #AI #SharingInPublic #LearningMindset
š—Ŗš—µš˜† š—”š—œ š—›š—®š—¹š—¹š˜‚š—°š—¶š—»š—®š˜š—²š˜€ š—®š—»š—± š—›š—¼š˜„ š—„š—”š—š š—™š—¼š—暝—°š—²š˜€ š—§š—暝˜‚š˜š—µ? Attended š—„š—”š—š š—˜š˜…š—½š—¹š—®š—¶š—»š—²š—±: š—§š—µš—² š—”š—暝—°š—µš—¶š˜š—²š—°š˜š˜‚š—暝—² š—•š—²š—µš—¶š—»š—± š—”š—“š—²š—»š˜š—¶š—° š—”š—œ š—¦š˜†š˜€š˜š—²š—ŗš˜€ lightening session hosted by Aishwarya and Arvind packed with practical use cases, architecture insights, and scalable design patterns for Agentic AI. š—§š—µš—² š—›š—®š—暝—± š—§š—暝˜‚š˜š—µ:,Your LLM will confidently sabotage you unless you anchor it reality. š—„š—²š—®š—¹-š—Ŗš—¼š—暝—¹š—± š—˜š˜…š—®š—ŗš—½š—¹š—²: Ask "š˜Šš˜¢š˜Æ š˜ š˜“š˜©š˜¢š˜³š˜¦ š˜·š˜¦š˜Æš˜„š˜°š˜³ š˜§š˜Ŗš˜­š˜¦š˜“?"ā†’ ChatGPT says "š˜ š˜¦š˜“, š˜©š˜¦š˜³š˜¦'š˜“ š˜©š˜°š˜ø" Your Policy: "š˜•š˜° š˜¦š˜¹š˜µš˜¦š˜³š˜Æš˜¢š˜­ š˜“š˜©š˜¢š˜³š˜Ŗš˜Æš˜Ø š˜øš˜Ŗš˜µš˜©š˜°š˜¶š˜µ š˜¢š˜±š˜±š˜³š˜°š˜·š˜¢š˜­" The Gap: Internet-trained LLMs don't know your rules. š˜Žš˜¦š˜®š˜Ŗš˜Æš˜Ŗ-generated session summary image from my notes along with Key Takeaways: š— š˜†š˜š—µ š—•š˜‚š˜€š˜š—²š—±: Bigger context windows don't replace retrieval š—Ÿš—¶š˜ƒš—² š—£š—暝—¼š—¼š—³: n8n demo turned a PDF into a cited chatbot instantly š—„š—²š—®š—¹š—¶š˜š˜† š—–š—µš—²š—°š—ø: Lag exists, RBAC needs work, data quality rules all š—•š—¼š˜š˜š—¼š—ŗ š—Ÿš—¶š—»š—²: Stop playing with consumer AI. Build retrieval pipelines that cite your truth or bet your business on fabricated answers. Big thanks to Aishwarya and Arvind for an highly informative session with live demo and Q&A. P.S: Personal views. #EnterpriseAI #RAG #AgenticAI #AIGovernance #AI #SharingInPublic #LearningMindset

RB
Raghavendra Bagalkoti

Building the Future of Finance with AI + Cloud | VP at Barclays | Simplifying Capital Markets | Thought Leader in Tech-Driven Finance

RAG is Dead!! Content window eliminates RAG!! Yesterday I attended the session ā€œRAG Explained: The Architecture Behind Agentic AI Systemsā€ on Maven, by Aishwarya Srinivasan & Arvind Narayanamurthy, that demystified so many myths about RAG, like the ones above šŸ‘† It was very insightful to learn LLMs alone are not enough. They are powerful reasoning engines, but they do not know your company data, internal documents, or real-time systems. Thatā€™s where RAG changes everything. Instead of asking the model to guess, a RAG system: 1ļøāƒ£ Retrieves relevant information 2ļøāƒ£ Injects it into the prompt 3ļøāƒ£ Lets the model reason over actual data In simple terms: LLM = šŸ§  Brain RAG = šŸ“š Memory Agents = šŸ¤– Workers Put them together and you get modern AI systems that can actually operate in the real world. šŸ‘ Huge thanks to the authors for making a complex topic clear, practical, and accessible. #ArtificialIntelligence #RAG #AgenticAI #AIArchitecture #GenerativeAI #LLM #AIEngineering
RAG is Dead!! Content window eliminates RAG!! Yesterday I attended the session ā€œRAG Explained: The Architecture Behind Agentic AI Systemsā€ on Maven, by Aishwarya Srinivasan & Arvind Narayanamurthy, that demystified so many myths about RAG, like the ones above šŸ‘† It was very insightful to learn LLMs alone are not enough. They are powerful reasoning engines, but they do not know your company data, internal documents, or real-time systems. Thatā€™s where RAG changes everything. Instead of asking the model to guess, a RAG system: 1ļøāƒ£ Retrieves relevant information 2ļøāƒ£ Injects it into the prompt 3ļøāƒ£ Lets the model reason over actual data In simple terms: LLM = šŸ§  Brain RAG = šŸ“š Memory Agents = šŸ¤– Workers Put them together and you get modern AI systems that can actually operate in the real world. šŸ‘ Huge thanks to the authors for making a complex topic clear, practical, and accessible. #ArtificialIntelligence #RAG #AgenticAI #AIArchitecture #GenerativeAI #LLM #AIEngineering

Jeel Jani

AI Product Manager | B2B SaaS | Rapid Prototyping (Cursor, Lovable) & LLM Prompt Engineering | YC Buildspace Alum

This weekend, I attended a RAG workshop by Aishwarya and Arvind - and I'm still thinking about it. Honestly, I went in expecting dense concepts, jargon, and that familiar feeling of nodding along while quietly being lost. What I got instead was clarity. What stood out wasn't just the content - it was how it was delivered. Complex ideas broken down with analogies that just *clicked*. The kind of teaching where you stop and think, "wait, that actually makes so much sense." RAG (Retrieval-Augmented Generation) is one of those topics that sounds intimidating until someone takes the time to meet you where you are. Aishwarya and Arvind did exactly that. Genuinely one of the best learning experiences I've had in a while.Also excited for upcoming workshops by The Gen Academy Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for making something powerful feel accessible. That's a rare skill. #RAG #MachineLearning #AI #LearningAndDevelopment #Workshop
This weekend, I attended a RAG workshop by Aishwarya and Arvind - and I'm still thinking about it. Honestly, I went in expecting dense concepts, jargon, and that familiar feeling of nodding along while quietly being lost. What I got instead was clarity. What stood out wasn't just the content - it was how it was delivered. Complex ideas broken down with analogies that just *clicked*. The kind of teaching where you stop and think, "wait, that actually makes so much sense." RAG (Retrieval-Augmented Generation) is one of those topics that sounds intimidating until someone takes the time to meet you where you are. Aishwarya and Arvind did exactly that. Genuinely one of the best learning experiences I've had in a while.Also excited for upcoming workshops by The Gen Academy Thank you Aishwarya Srinivasan and Arvind Narayanamurthy for making something powerful feel accessible. That's a rare skill. #RAG #MachineLearning #AI #LearningAndDevelopment #Workshop

Arjun Rajan

Senior System Analyst at CSC

Grateful for an incredibly insightful session on Claude Cowork and Claude Code! šŸ™Œ A big thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and the entire team at The Gen Academy for putting together such a well-structured and engaging session. The depth of knowledge, practical insights, and clarity in explaining real-world applications made the learning experience truly valuable. Itā€™s sessions like these that help bridge the gap between concepts and implementation. Looking forward to applying these learnings and exploring more in this space! #AI #Claude #Learning #TechCommunity #GenAcademy #DeveloperGrowth
Grateful for an incredibly insightful session on Claude Cowork and Claude Code! šŸ™Œ A big thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and the entire team at The Gen Academy for putting together such a well-structured and engaging session. The depth of knowledge, practical insights, and clarity in explaining real-world applications made the learning experience truly valuable. Itā€™s sessions like these that help bridge the gap between concepts and implementation. Looking forward to applying these learnings and exploring more in this space! #AI #Claude #Learning #TechCommunity #GenAcademy #DeveloperGrowth

Y
Yogitha R

QA Engineer at Aptean | AI-Native in the Making | LLMs Ā· Agentic AI Ā· GenAI Workflows | Turning 10x Productivity into Reality

This Saturday had a different kind of energy. No plans. Just curiosity - and a session I had been waiting for all week. "The AI PM Playbook: Building AI Products at Scale" by MAVEN. And honestly? It delivered more than I expected. What hit me the most wasn't just the content. It was watching Aishwarya Srinivasan walk through her entire AI stack live - Claude for PRDs and stakeholder narratives, Claude Code for prototyping agents, Lovable for building full apps from prompts without a dev bottleneck. That last one sat with me. No dev bottleneck. As someone who works in a product-based company and has spent time deeply understanding how products are built, tested, and shipped - this felt like a signal I couldn't ignore. The gap between "having product intuition" and "building something with it" is getting smaller every single day. And the people who figure out how to combine product thinking with AI agent workflows? They're going to move at a speed that's just different. I left the session with 3 pages of notes, 2 new tools to try, and 1 very clear thought: The next thing I build won't just use AI. It'll be powered by agents that think, act, and ship alongside me. Massive thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and Shyamala Prayaga for putting together something real. No fluff. Just depth. If you missed it , keep an eye on MAVEN. Worth every SaturdayšŸ™Œ #AIProductManagement #BuildingWithAI #AIAgents #ProductThinking #MAVENlearning #ClaudeAI #GenAI #AIWorkflow #WeekendLearning #FutureOfProduct#AIProductManagement #BuildingWithAI #AIAgents #ProductThinking #MAVENlearning #ClaudeAI #GenAI #AIWorkflow #WeekendLearning #FutureOfProduct
This Saturday had a different kind of energy. No plans. Just curiosity - and a session I had been waiting for all week. "The AI PM Playbook: Building AI Products at Scale" by MAVEN. And honestly? It delivered more than I expected. What hit me the most wasn't just the content. It was watching Aishwarya Srinivasan walk through her entire AI stack live - Claude for PRDs and stakeholder narratives, Claude Code for prototyping agents, Lovable for building full apps from prompts without a dev bottleneck. That last one sat with me. No dev bottleneck. As someone who works in a product-based company and has spent time deeply understanding how products are built, tested, and shipped - this felt like a signal I couldn't ignore. The gap between "having product intuition" and "building something with it" is getting smaller every single day. And the people who figure out how to combine product thinking with AI agent workflows? They're going to move at a speed that's just different. I left the session with 3 pages of notes, 2 new tools to try, and 1 very clear thought: The next thing I build won't just use AI. It'll be powered by agents that think, act, and ship alongside me. Massive thank you to Aishwarya Srinivasan, Arvind Narayanamurthy, and Shyamala Prayaga for putting together something real. No fluff. Just depth. If you missed it , keep an eye on MAVEN. Worth every SaturdayšŸ™Œ #AIProductManagement #BuildingWithAI #AIAgents #ProductThinking #MAVENlearning #ClaudeAI #GenAI #AIWorkflow #WeekendLearning #FutureOfProduct#AIProductManagement #BuildingWithAI #AIAgents #ProductThinking #MAVENlearning #ClaudeAI #GenAI #AIWorkflow #WeekendLearning #FutureOfProduct

Karthik Malaiyappan

Data | AI | Agentic | FinTech

Thank you so much Aishwarya Srinivasan and Arvind Narayanamurthy for hosting Session on Building AI Products at Scale. Appreciate the effort you put out to brought the knowledge from core of its industry to the people.Looking for more such sessions. Honored to learn some perspective from its core.Shyamala Prayaga you were just amazing throughout the session.The explanation you gave for all the questions is quite understandable in simple words. AI is not thread itself it depends on how we engage it with products we builtāœØāœØ #AI #Agentic #Saturdaylearning #Products
Thank you so much Aishwarya Srinivasan and Arvind Narayanamurthy for hosting Session on Building AI Products at Scale. Appreciate the effort you put out to brought the knowledge from core of its industry to the people.Looking for more such sessions. Honored to learn some perspective from its core.Shyamala Prayaga you were just amazing throughout the session.The explanation you gave for all the questions is quite understandable in simple words. AI is not thread itself it depends on how we engage it with products we builtāœØāœØ #AI #Agentic #Saturdaylearning #Products

Mahaveer Satra

Solutions Architect | Digital Twins | Building Agentic AI Systems

I spent 3 months building a multi-agent system that does 2-4 hours of AE work in ~4 minutes.   4 LangGraph agents. Cited sales strategy. Production-grade.   3 things production taught me:   -> NAIVE RAG FAILS HARD "GNC toolboxes" returned zero semantic similarity to "Sensor Fusion and Tracking Toolbox." Vector-only search = wrong products in front of customers.   Fixed it with Hybrid RAG (thanks Aishwarya Srinivasan): BM25 + vector + cross-encoder reranking. Retrieval quality is the ceilingā€”everything downstream inherits its failures.   -> YOU CAN'T TUNE WITHOUT EVALS Like adjusting PID gains without feedback means you're flying blind.   Built Claude-as-judge with 4 metrics + 9 deterministic checks. Went from 3.0 to 4.2 quality score across 3 iterations.   -> LLM COST IS A ROUTING PROBLEM Ollama ($0) ā†’ Groq 8B ā†’ Groq 70B dispatched by task complexity. Prompt caching cuts 90% of token costs. MIT licensed. If any of this saves you a debugging session, fork it, build on it. Repo Link in comments.   #AgenticAI #LangGraph #RAG
I spent 3 months building a multi-agent system that does 2-4 hours of AE work in ~4 minutes.   4 LangGraph agents. Cited sales strategy. Production-grade.   3 things production taught me:   -> NAIVE RAG FAILS HARD "GNC toolboxes" returned zero semantic similarity to "Sensor Fusion and Tracking Toolbox." Vector-only search = wrong products in front of customers.   Fixed it with Hybrid RAG (thanks Aishwarya Srinivasan): BM25 + vector + cross-encoder reranking. Retrieval quality is the ceilingā€”everything downstream inherits its failures.   -> YOU CAN'T TUNE WITHOUT EVALS Like adjusting PID gains without feedback means you're flying blind.   Built Claude-as-judge with 4 metrics + 9 deterministic checks. Went from 3.0 to 4.2 quality score across 3 iterations.   -> LLM COST IS A ROUTING PROBLEM Ollama ($0) ā†’ Groq 8B ā†’ Groq 70B dispatched by task complexity. Prompt caching cuts 90% of token costs. MIT licensed. If any of this saves you a debugging session, fork it, build on it. Repo Link in comments.   #AgenticAI #LangGraph #RAG

Lijo Joseph

PMP Certified Project Manager | Senior Business Analyst | Scrum Master | Enterprise Delivery, Change, Governance, Digital Transformation | AWS Cloud | Lean Six Sigma | BFSI | UAE Resident Visa valid until 2036

Most AI answers š˜“š˜°š˜¶š˜Æš˜„ š˜³š˜Ŗš˜Øš˜©š˜µā€¦ but are they actually correct? One of the challenges with LLMs becomes very clear in enterprise settings. LLMs like ChatGPT, Gemini, or Claude are trained on š—¶š—»š˜š—²š—暝—»š—²š˜ š—±š—®š˜š—®, not your companyā€™s internal policies, processes, or documents. So when someone asks: ā€œš˜Šš˜¢š˜Æ š˜ š˜“š˜©š˜¢š˜³š˜¦ š˜¢ š˜¤š˜­š˜Ŗš˜¦š˜Æš˜µ š˜§š˜Ŗš˜­š˜¦ š˜øš˜Ŗš˜µš˜© š˜¢ š˜·š˜¦š˜Æš˜„š˜°š˜³?ā€ The response may sound confident and logicalā€¦ but it could still be š—ŗš—¶š˜€š—®š—¹š—¶š—“š—»š—²š—± š˜„š—¶š˜š—µ š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š—½š—¼š—¹š—¶š—°š˜†. š—§š—µš—¶š˜€ š—¶š˜€ š˜„š—µš—²š—暝—² š—„š—”š—š (š—„š—²š˜š—暝—¶š—²š˜ƒš—®š—¹ š—”š˜‚š—“š—ŗš—²š—»š˜š—²š—± š—šš—²š—»š—²š—暝—®š˜š—¶š—¼š—») š—°š—¼š—ŗš—²š˜€ š—¶š—» RAG connects LLMs to š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š—øš—»š—¼š˜„š—¹š—²š—±š—“š—² š˜š—µš—²š˜† š˜„š—²š—暝—² š—»š—²š˜ƒš—²š—æ š˜š—暝—®š—¶š—»š—²š—± š—¼š—». Instead of relying only on pre-trained knowledge, it:  1. Breaks documents into š—°š—µš˜‚š—»š—øš˜€  2. Converts them into š—²š—ŗš—Æš—²š—±š—±š—¶š—»š—“š˜€ (numerical representations of text)  3. Stores them in a š˜ƒš—²š—°š˜š—¼š—æ š—±š—®š˜š—®š—Æš—®š˜€š—²  4. Uses š—°š—¼š˜€š—¶š—»š—² š˜€š—¶š—ŗš—¶š—¹š—®š—暝—¶š˜š˜†(=0.97) to retrieve the most relevant context  5. Feeds that context to the LLM to generate grounded responses š—–š—µš˜‚š—»š—øš—¶š—»š—“ š—¶š˜€ š—ŗš—¼š—暝—² š—¶š—ŗš—½š—¼š—暝˜š—®š—»š˜ š˜š—µš—®š—» š—¶š˜ š—¹š—¼š—¼š—øš˜€ The way you split data directly impacts the quality of retrieval. Common chunking strategies include:  ā€¢ Fixed-length chunking  ā€¢ Sentence-based chunking  ā€¢ Paragraph-based chunking  ā€¢ Sliding window chunking  ā€¢ Semantic chunking  ā€¢ Recursive chunking Better chunking ā†’ Better retrieval ā†’ Better answers And more importantly: š—•š—²š˜š˜š—²š—æ š—®š—»š˜€š˜„š—²š—暝˜€ š˜„š—¶š˜š—µ š—°š—¶š˜š—®š˜š—¶š—¼š—»š˜€ Which means the response is not just helpful, but š˜š—暝—®š—°š—²š—®š—Æš—¹š—²:  ā€¢ You can see where the answer came from  ā€¢ You can refer back to the exact company document  ā€¢ The answer has real grounding, not just confidence Thatā€™s what gives AI š—°š—暝—²š—±š—¶š—Æš—¶š—¹š—¶š˜š˜† š—¶š—» š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š˜‚š˜€š—². āš–ļøš—„š—”š—š š˜ƒš˜€ š—™š—¶š—»š—²-š˜š˜‚š—»š—¶š—»š—“ š—™š—¶š—»š—²-š˜š˜‚š—»š—¶š—»š—“ ā†’ Good for behavior, structure, and low latency š—„š—”š—š ā†’ Better for dynamic, real-time, and private enterprise data For most enterprise use cases, RAG is the most effective starting point for improving LLM responses. š— š˜† š˜š—®š—øš—²š—®š˜„š—®š˜† AI is not just about generating answers. Itā€™s about:  ā€¢ Retrieving the right context  ā€¢ Grounding responses in real data  ā€¢ Making answers š˜ƒš—²š—暝—¶š—³š—¶š—®š—Æš—¹š—², š—»š—¼š˜ š—·š˜‚š˜€š˜ š—°š—¼š—»š˜ƒš—¶š—»š—°š—¶š—»š—“ A quick shoutout to Arvind Narayanamurthy & Aishwarya Srinivasan from The Gen Academy for this weekend learning.
Most AI answers š˜“š˜°š˜¶š˜Æš˜„ š˜³š˜Ŗš˜Øš˜©š˜µā€¦ but are they actually correct? One of the challenges with LLMs becomes very clear in enterprise settings. LLMs like ChatGPT, Gemini, or Claude are trained on š—¶š—»š˜š—²š—暝—»š—²š˜ š—±š—®š˜š—®, not your companyā€™s internal policies, processes, or documents. So when someone asks: ā€œš˜Šš˜¢š˜Æ š˜ š˜“š˜©š˜¢š˜³š˜¦ š˜¢ š˜¤š˜­š˜Ŗš˜¦š˜Æš˜µ š˜§š˜Ŗš˜­š˜¦ š˜øš˜Ŗš˜µš˜© š˜¢ š˜·š˜¦š˜Æš˜„š˜°š˜³?ā€ The response may sound confident and logicalā€¦ but it could still be š—ŗš—¶š˜€š—®š—¹š—¶š—“š—»š—²š—± š˜„š—¶š˜š—µ š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š—½š—¼š—¹š—¶š—°š˜†. š—§š—µš—¶š˜€ š—¶š˜€ š˜„š—µš—²š—暝—² š—„š—”š—š (š—„š—²š˜š—暝—¶š—²š˜ƒš—®š—¹ š—”š˜‚š—“š—ŗš—²š—»š˜š—²š—± š—šš—²š—»š—²š—暝—®š˜š—¶š—¼š—») š—°š—¼š—ŗš—²š˜€ š—¶š—» RAG connects LLMs to š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š—øš—»š—¼š˜„š—¹š—²š—±š—“š—² š˜š—µš—²š˜† š˜„š—²š—暝—² š—»š—²š˜ƒš—²š—æ š˜š—暝—®š—¶š—»š—²š—± š—¼š—». Instead of relying only on pre-trained knowledge, it:  1. Breaks documents into š—°š—µš˜‚š—»š—øš˜€  2. Converts them into š—²š—ŗš—Æš—²š—±š—±š—¶š—»š—“š˜€ (numerical representations of text)  3. Stores them in a š˜ƒš—²š—°š˜š—¼š—æ š—±š—®š˜š—®š—Æš—®š˜€š—²  4. Uses š—°š—¼š˜€š—¶š—»š—² š˜€š—¶š—ŗš—¶š—¹š—®š—暝—¶š˜š˜†(=0.97) to retrieve the most relevant context  5. Feeds that context to the LLM to generate grounded responses š—–š—µš˜‚š—»š—øš—¶š—»š—“ š—¶š˜€ š—ŗš—¼š—暝—² š—¶š—ŗš—½š—¼š—暝˜š—®š—»š˜ š˜š—µš—®š—» š—¶š˜ š—¹š—¼š—¼š—øš˜€ The way you split data directly impacts the quality of retrieval. Common chunking strategies include:  ā€¢ Fixed-length chunking  ā€¢ Sentence-based chunking  ā€¢ Paragraph-based chunking  ā€¢ Sliding window chunking  ā€¢ Semantic chunking  ā€¢ Recursive chunking Better chunking ā†’ Better retrieval ā†’ Better answers And more importantly: š—•š—²š˜š˜š—²š—æ š—®š—»š˜€š˜„š—²š—暝˜€ š˜„š—¶š˜š—µ š—°š—¶š˜š—®š˜š—¶š—¼š—»š˜€ Which means the response is not just helpful, but š˜š—暝—®š—°š—²š—®š—Æš—¹š—²:  ā€¢ You can see where the answer came from  ā€¢ You can refer back to the exact company document  ā€¢ The answer has real grounding, not just confidence Thatā€™s what gives AI š—°š—暝—²š—±š—¶š—Æš—¶š—¹š—¶š˜š˜† š—¶š—» š—²š—»š˜š—²š—暝—½š—暝—¶š˜€š—² š˜‚š˜€š—². āš–ļøš—„š—”š—š š˜ƒš˜€ š—™š—¶š—»š—²-š˜š˜‚š—»š—¶š—»š—“ š—™š—¶š—»š—²-š˜š˜‚š—»š—¶š—»š—“ ā†’ Good for behavior, structure, and low latency š—„š—”š—š ā†’ Better for dynamic, real-time, and private enterprise data For most enterprise use cases, RAG is the most effective starting point for improving LLM responses. š— š˜† š˜š—®š—øš—²š—®š˜„š—®š˜† AI is not just about generating answers. Itā€™s about:  ā€¢ Retrieving the right context  ā€¢ Grounding responses in real data  ā€¢ Making answers š˜ƒš—²š—暝—¶š—³š—¶š—®š—Æš—¹š—², š—»š—¼š˜ š—·š˜‚š˜€š˜ š—°š—¼š—»š˜ƒš—¶š—»š—°š—¶š—»š—“ A quick shoutout to Arvind Narayanamurthy & Aishwarya Srinivasan from The Gen Academy for this weekend learning.

Michael Galarnyk

Research Scientist Intern @ Adobe | Dell Pro Precision AmbassadorĀ |Ā ML Ph.D. @ Georgia Tech

I ran 7 RAG architectures side by side on the same questions to see what actually matters. The setup comes from a deep dive by a friend of mine, Vidhyakshaya Kannan, on The Gen Academy which breaks down these RAG architectures and their tradeoffs. A few things that stood out: - Most architectures land in a similar accuracy range - Agentic RAG performs best, but costs ~4.8Ɨ more per correct answer - Self-RAG underperforms Naive RAG. More complexity, worse results It is easy to over-engineer RAG systems. Running them side by side makes that tradeoff very clear. Iā€™ll drop the blog post and code in the comments. How are you deciding between RAG architectures in practice? Optimizing for accuracy, cost, or something else? #MachineLearning #ArtificialIntelligence #DellTech #DellProPrecision #NVIDIA
I ran 7 RAG architectures side by side on the same questions to see what actually matters. The setup comes from a deep dive by a friend of mine, Vidhyakshaya Kannan, on The Gen Academy which breaks down these RAG architectures and their tradeoffs. A few things that stood out: - Most architectures land in a similar accuracy range - Agentic RAG performs best, but costs ~4.8Ɨ more per correct answer - Self-RAG underperforms Naive RAG. More complexity, worse results It is easy to over-engineer RAG systems. Running them side by side makes that tradeoff very clear. Iā€™ll drop the blog post and code in the comments. How are you deciding between RAG architectures in practice? Optimizing for accuracy, cost, or something else? #MachineLearning #ArtificialIntelligence #DellTech #DellProPrecision #NVIDIA

Praneeta Pericherla

Incoming MS-DAS student @ Carnegie Mellon University

Most AI PMs are managing probabilistic systems with deterministic tools. That's why AI products quietly rot after launch. In traditional software: a button either works or it doesn't. Bugs are reproducible. If it passes staging, it passes prod. In AI: the "same" input produces different outputs. A prompt that works on 10 examples fails on the 11th. Features don't ship as "works / doesn't work", they ship with a distribution of quality. And most PMs have no idea how to measure that distribution. I mapped out the full AI PM tooling stack for 2026, the 6 layers every AI PM needs to know, the category leaders in each, and the judgment calls that actually matter: ā†’ Evaluation (where quality is defined) ā†’ Observability (seeing what actually shipped) ā†’ Prompt management (governing changes safely) ā†’ Prototyping (Bolt, Lovable, v0, Cursor) ā†’ Experimentation (shipping to users safely) ā†’ Customer feedback (keeping evals honest) The biggest mistake AI PMs make? Adopting an eval tool and never building a real dataset. Treating the tool as the deliverable instead of the scaffolding. Inspired by the Lightning Session on The AI PM Playbook by The Gen Academy, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy, with guest speaker Shyamala Prayaga. Full breakdown on Substack, link in comments šŸ‘‡
Most AI PMs are managing probabilistic systems with deterministic tools. That's why AI products quietly rot after launch. In traditional software: a button either works or it doesn't. Bugs are reproducible. If it passes staging, it passes prod. In AI: the "same" input produces different outputs. A prompt that works on 10 examples fails on the 11th. Features don't ship as "works / doesn't work", they ship with a distribution of quality. And most PMs have no idea how to measure that distribution. I mapped out the full AI PM tooling stack for 2026, the 6 layers every AI PM needs to know, the category leaders in each, and the judgment calls that actually matter: ā†’ Evaluation (where quality is defined) ā†’ Observability (seeing what actually shipped) ā†’ Prompt management (governing changes safely) ā†’ Prototyping (Bolt, Lovable, v0, Cursor) ā†’ Experimentation (shipping to users safely) ā†’ Customer feedback (keeping evals honest) The biggest mistake AI PMs make? Adopting an eval tool and never building a real dataset. Treating the tool as the deliverable instead of the scaffolding. Inspired by the Lightning Session on The AI PM Playbook by The Gen Academy, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy, with guest speaker Shyamala Prayaga. Full breakdown on Substack, link in comments šŸ‘‡

Rachana NV

Software Engineer | .NET | Distributed Systems | Building User-Centric Products | Ex-CommBank | Ex-Hyland R&D

Yesterday, I attended the AI PM Playbook by The Gen Academy, featuring Shyamala Prayaga (Senior AI Product Manager at NVIDIA). You might wonder, whatā€™s a software engineer doing in an AI PM session? Because Iā€™ve seen how often we jump straight into buildingā€¦ and only later realize we shouldā€™ve asked better questions first. Working closely with PMs, Iā€™ve started valuing that thought process more, how ideas are shaped before they become features. Thatā€™s what pulled me into this session. One line from the session that really stayed with me: šŸ‘‰ Donā€™t fall for the ā€œshiny object syndromeā€ in AI. Itā€™s tempting to add AI everywhere, but the real focus should be on: ā€¢ Solving the right problem (not just using AI for the sake of it) ā€¢ Defining metrics & evals upfront - ā€œbuild evals before building featuresā€ ā€¢ Thinking about adoption - token economics, context, and real user value A good reminder that building AI products, at any level, needs clarity, intent, and responsibility, not just hype. #AI #ProductManagement #SoftwareEngineering #GenAI #Learning
Yesterday, I attended the AI PM Playbook by The Gen Academy, featuring Shyamala Prayaga (Senior AI Product Manager at NVIDIA). You might wonder, whatā€™s a software engineer doing in an AI PM session? Because Iā€™ve seen how often we jump straight into buildingā€¦ and only later realize we shouldā€™ve asked better questions first. Working closely with PMs, Iā€™ve started valuing that thought process more, how ideas are shaped before they become features. Thatā€™s what pulled me into this session. One line from the session that really stayed with me: šŸ‘‰ Donā€™t fall for the ā€œshiny object syndromeā€ in AI. Itā€™s tempting to add AI everywhere, but the real focus should be on: ā€¢ Solving the right problem (not just using AI for the sake of it) ā€¢ Defining metrics & evals upfront - ā€œbuild evals before building featuresā€ ā€¢ Thinking about adoption - token economics, context, and real user value A good reminder that building AI products, at any level, needs clarity, intent, and responsibility, not just hype. #AI #ProductManagement #SoftwareEngineering #GenAI #Learning

Pawan Simha

Student @ SNPSU | Gemini Certified | Aspiring AI Product Manager | Python ā€¢ Machine Learning ā€¢ Web Development | CR & HRD Coordinator

AI products donā€™t fail because of bad models. They fail because of bad decisions. Over the weekend, I attended the Lightning Lesson on The AI PM Playbook, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy (Maven Ɨ The Gen Academy), with Shyamala Prayaga (NVIDIA). It exposed a gap in how I was thinking about AI systems. I was focused on capabilities. This shifted me toward accountability. The MAP Framework (Model ā†’ Augment ā†’ Program) stood outā€”not just as a technique, but as a decision filter: 1. When should AI generate? 2. When should it assist? 3. And when should it stay out entirely? This matters because AI is non-deterministic. Treat it like traditional software, and you end up with "silent failures"ā€”systems that appear correct until they break user trust. The shift for me: As someone with a technical background, I used to ask: ā€œWhat can this model do?ā€ Now, as an aspiring AI PM, I ask: ā€œWhat should the system be responsible for?ā€ Iā€™m now applying this product-first mindset to my own projectsā€”moving beyond demos toward reliable systems at scale. For AI builders: whatā€™s one decision that becomes harder when moving from demo to production? #ArtificialIntelligence #NVIDIA #AIProductManagement #GenerativeAI #SystemDesign #BuildInPublic #AIEngineering #MachineLearning #ProductStrategy #TheGenAcademy
AI products donā€™t fail because of bad models. They fail because of bad decisions. Over the weekend, I attended the Lightning Lesson on The AI PM Playbook, hosted by Aishwarya Srinivasan and Arvind Narayanamurthy (Maven Ɨ The Gen Academy), with Shyamala Prayaga (NVIDIA). It exposed a gap in how I was thinking about AI systems. I was focused on capabilities. This shifted me toward accountability. The MAP Framework (Model ā†’ Augment ā†’ Program) stood outā€”not just as a technique, but as a decision filter: 1. When should AI generate? 2. When should it assist? 3. And when should it stay out entirely? This matters because AI is non-deterministic. Treat it like traditional software, and you end up with "silent failures"ā€”systems that appear correct until they break user trust. The shift for me: As someone with a technical background, I used to ask: ā€œWhat can this model do?ā€ Now, as an aspiring AI PM, I ask: ā€œWhat should the system be responsible for?ā€ Iā€™m now applying this product-first mindset to my own projectsā€”moving beyond demos toward reliable systems at scale. For AI builders: whatā€™s one decision that becomes harder when moving from demo to production? #ArtificialIntelligence #NVIDIA #AIProductManagement #GenerativeAI #SystemDesign #BuildInPublic #AIEngineering #MachineLearning #ProductStrategy #TheGenAcademy