Building Agentic AI Applications with a Problem-First Approach Reviews on Testimonial.to

Aviral Gupta

The last few weeks were miserable. A grueling, hands-on AI course on top of my usual fare. This was the kind where you're not vibe coding but actually writing code at 11pm wondering if any of it is landing. Classic type 2 fun: awful while it's happening, and now I miss it. On the final day, Kiriti & Aishwarya told us something that had nothing to do with how it felt. They'd been running what basketball coaches call a head fake on us the whole time: teaching a real skill while you think you're learning something else entirely. Here's what that means: 1. Meta-learn, don't just learn. This can feel oddly frustrating in the moment. My daughter can vouch. I rarely give a straight answer on purpose; most questions get a question back to make you dig deeper. Aish and Kiriti did something like this, just with more restraint than I apparently have at home. 2. You can't catch up, and that's fine. Especially true in the content-overload world we're all living in right now. Aish and Kiriti shared something simple: nobody's actually caught up, so stop trying and go deep on your own lane instead. I'm going to try. But FOMO and YOLO work against each other and I've never been great at picking a side. 3. Solve it the simplest way possible. We need to decouple complexity and smarts. Too often we feel smart by cramming in every concept and walking in with the "perfect" solution already built. In reality it's baby steps: crawl, walk, run. FYL has massive aspirations to build a store for each user. I still start by asking: what events already exist today, and is anyone acting on them — or are they just piling up in a warehouse? 4. Design thinking gets more expensive as implementation gets cheaper. Another way to put it: judgement is what you get paid for. Product taste, design sense, whatever you want to call it: don't outsource it. Anyone can wire up a recommendation widget; the judgement is knowing what to personalize and when. 5. “It depends”, beats hot takes. Every other week it's a new death notice: PM is dead, design is dead, RAG is dead, this framework is the only one that matters now. Breathe. Dig in. Ask what has to be true for that claim to hold before you let it change what you do. 6. You belong here. I hadn't written code in 20 years. I resisted the pull toward Langflow and wrote raw Python instead. Felt empowering. Today, the tools meet you where you are now, low-code or not. I now spend more time in a terminal window than in a Google Doc. If you'd told me that a year ago, I wouldn't have believed it. I was overwhelmed with Agent this and Agent that. The course (lnkd.in/gegkCBFH) demystified it by not only sharing core concepts but re-emphasizing a problem-first approach: what are you actually solving, does an agent even help, what's the simplest version that works first. That demystifying, more than any framework, is what I'm taking with me.

The last few weeks were miserable. A grueling, hands-on AI course on top of my usual fare. This was the kind where you're not vibe coding but actually writing code at 11pm wondering if any of it is landing. Classic type 2 fun: awful while it's happening, and now I miss it. On the final day, Kiriti & Aishwarya told us something that had nothing to do with how it felt. They'd been running what basketball coaches call a head fake on us the whole time: teaching a real skill while you think you're learning something else entirely. Here's what that means: 1. Meta-learn, don't just learn. This can feel oddly frustrating in the moment. My daughter can vouch. I rarely give a straight answer on purpose; most questions get a question back to make you dig deeper. Aish and Kiriti did something like this, just with more restraint than I apparently have at home. 2. You can't catch up, and that's fine. Especially true in the content-overload world we're all living in right now. Aish and Kiriti shared something simple: nobody's actually caught up, so stop trying and go deep on your own lane instead. I'm going to try. But FOMO and YOLO work against each other and I've never been great at picking a side. 3. Solve it the simplest way possible. We need to decouple complexity and smarts. Too often we feel smart by cramming in every concept and walking in with the "perfect" solution already built. In reality it's baby steps: crawl, walk, run. FYL has massive aspirations to build a store for each user. I still start by asking: what events already exist today, and is anyone acting on them — or are they just piling up in a warehouse? 4. Design thinking gets more expensive as implementation gets cheaper. Another way to put it: judgement is what you get paid for. Product taste, design sense, whatever you want to call it: don't outsource it. Anyone can wire up a recommendation widget; the judgement is knowing what to personalize and when. 5. “It depends”, beats hot takes. Every other week it's a new death notice: PM is dead, design is dead, RAG is dead, this framework is the only one that matters now. Breathe. Dig in. Ask what has to be true for that claim to hold before you let it change what you do. 6. You belong here. I hadn't written code in 20 years. I resisted the pull toward Langflow and wrote raw Python instead. Felt empowering. Today, the tools meet you where you are now, low-code or not. I now spend more time in a terminal window than in a Google Doc. If you'd told me that a year ago, I wouldn't have believed it. I was overwhelmed with Agent this and Agent that. The course (lnkd.in/gegkCBFH) demystified it by not only sharing core concepts but re-emphasizing a problem-first approach: what are you actually solving, does an agent even help, what's the simplest version that works first. That demystifying, more than any framework, is what I'm taking with me.

Sep 16, 2026

Bharath Devanathan

I am usually quite wary of online AI courses - most of them are hype and give you this impression that you need to “catch up” with the rapid pace of advancement in the AI world. I therefore wanted to carefully evaluate the options, separate the signal from the noise and choose a course that teaches from first principles. I just finished (lnkd.in/gNGYw_6i) Building Agentic AI Applications with a Problem-First Approach. This course is taught by Aishwarya Naresh Reganti and Kiriti Badam - both of whom have significant experience in implementing and deploying enterprise grade solutions that have had significant impact across F500 organizations. They are also at the frontier of AI development themselves. I also really liked that the course stuck to software engineering first principles thinking - understand the domain, understand the use case and choose the best tools for the job. Just because AI makes things easy to build, building is not the first step. Determine what kind of AI architecture (along with test cases) will help derive the best solutions, map it out and then think of implementation. Also consciously choose the tradeoffs (agency vs control, latency vs cost/performance) and don’t let an agent do that for you. I learnt quite a bit about what happens behind the scenes with AI tools, how one can tune them to fit a particular use case and what are the ways to combine the toolsets available. I was able to bring down the cost of one implementation by an order of magnitude, without compromising performance. If you want to understand the inner workings of tools like Claude Code and want to build enterprise AI applications, this is a good course to take.

Sep 16, 2026

Gagandeep Pahwa

A bit of a late post, but have been busy absorbing all the learnings! I recently completed the 'Building Agentic AI Applications with a Problem-First Approach' course, led by the brilliant Aishwarya Naresh Reganti and Kiriti Badam! While it’s impossible to distill everything into a single post, here is my shot at sharing my top takeaways with respect to the core technical concepts and the invaluable 'meta-learnings'. 🛠️ Top 5 Concepts That Changed My Perspective: 1️⃣ Problem-First & Eval-First Design: Yes, evals matter from day one! Building iteratively is the only way. 2️⃣ Open-Source vs. Closed-Source: Nuanced frameworks on exactly when to use which models. 3️⃣ Prompt Engineering & Meta Prompting: Taking true control of model behavior. 4️⃣ Agentic Architecture: Deep dives into RAG, agentic retrieval, and memory. 5️⃣ Agency vs. Control: Knowing when to give an agent autonomy vs. when to build a deterministic workflow (multiple levels of this!!) 💡 3 "Meta-Learnings" (or "Head-Fakes" as Aish & Kiriti call them): 🔹 Separating the hype from reality: This is crucial not just in AI, but in navigating any new technology. 🔹 Real AI is "boring": Working systems are hard and complex. Most aren't "news-worthy" and won't get you viral LinkedIn likes. 🔹 The Art of Teaching that I will use to share my learning: Don't cater to the lowest common denominator. Challenge yourself to deliver high-signal learnings in the most consumable way possible. I am also incredibly proud to share my Certificate of Completion, alongside earning the Top Contributor Recognition and finishing as Runner-Up for the Capstone Project! 🏆 A massive shoutout to my fantastic capstone teammates for the collaboration: Gaurav Pahuja Tanya Shukla Barnali Patnaik Rohan Saxena! Grateful for the learning, and excited to get back to building! 🚀

Sep 16, 2026

Avinash Kumar

I recently completed the 5-week Maven course on “Building Agentic AI Applications with a Problem-First Approach”. - lnkd.in/drxqnCp9 What I liked most was that the course was not about simply building something with an LLM. It made me think more deeply about how to design AI systems that work reliably in real-world scenarios. We covered a wide range of topics, including deterministic vs. non-deterministic agents, token usage and context management, memory, different RAG approaches, MCP, multi-agent systems, HITL, observability, and evals. The section on evals and failure analysis was particularly useful. It reinforced the idea that building an agent is only the beginning. We need ways to measure its behaviour, understand where it fails, make improvements, and continuously test those changes. Thanks, Aishwarya Naresh Reganti and Kiriti Badam, for putting together such a practical course and for sharing your experience throughout the journey.

Sep 16, 2026

Srini Rangaswamy

With so much noise in the AI space and many building agents without first asking why we need one or what problem we are actually solving, it is refreshing to attend the "Building AI Apps with a Problem-First Approach" course. What I liked the most is the focus on the problem first and technology second! With AI being non-deterministic and requiring ongoing “behavior calibration”, the right solution approach is to incrementally introduce AI rather than rushing into using a fancy AI framework. This idea of incremental infusion of AI manifested itself in the capstone project where the teams presented agentic solution design across three incremental iterations. As building becomes easier it is important to take the time to design where intelligence should be applied and that idea is hammered home throughout the course. Thank you Aishwarya Naresh Reganti, Kiriti Badam, and team for this course! Check out the course link in comment.

Sep 16, 2026

Nilesh Shrimali

I just wrapped up Problem First AI, a 5-week Maven course by Aishwarya Naresh Reganti and Kiriti Badam Biggest takeaway: Model capability alone does not determine Agentic AI success. It is just a smaller piece of Agent Harness which is like a "mini OS" that works in conjunction with the model. This course really encourages taking the Problem First approach as we build out Enterprise AI systems. It covers implementation of agentic workflows using LangGraph along with Observability tools like Arize. It goes into details of RAG techniques as well as challenges with diff approaches and emphasizes eval-driven development. I would definitely recommend this to anyone who is starting their journey with AI Engineering.

Sep 16, 2026

Akash Jain

Just completed an intense and incredibly practical journey into Agentic AI at LevelUp Labs. The biggest takeaway for me? Building an AI agent is the easy part. Building one that you can measure, optimise, explain and trust is the real challenge. For the capstone, we built QuickBrief - an AI analyst that takes a business question in natural language and turns it into an evidence-backed, decision-ready analysis. Not just text-to-SQL, but the ability to investigate why something happened through root-cause analysis. What made the experience particularly valuable was evolving the system through multiple levels of agency and seeing firsthand what each architectural decision really costs. A few learnings I'll take away from the course: 🔹 Performance first, optimisation second. Get the system correct and measurable before making it cheaper or faster. Otherwise, you're just optimising the wrong thing. 🔹 More sophistication doesn't necessarily mean more value. Autonomous or multi-agent architectures can look impressive, but the right amount of agency matters more than maximum agency. 🔹 Start with a workflow. A deterministic workflow is easier to understand, test and govern. It provides a reliable foundation before introducing autonomy where it actually adds value. 🔹 Context engineering matters. More information isn't necessarily better. Structuring the right context can be far more effective than giving an LLM access to everything. 🔹 Modularity makes Evals meaningful. Atomic, testable components make it possible to understand where something went wrong rather than simply knowing the final answer was wrong. 🔹 Observability changes how you build AI systems. Tracing isn't just about monitoring production. During our capstone, traces helped us identify real bottlenecks and failures, allowing us to optimise specific components rather than guessing. Perhaps my biggest takeaway: Don't build an agent because you can. Build the simplest system that reliably solves the problem and earn the right to add autonomy. A huge thank you to Aishwarya Naresh Reganti, Kiriti Badam, Ravi Yenduri and the entire team for making the course so grounded in real design principles and trade-offs. Special shout-out to my truly global capstone team Deepam Agarwal, Abhishek Agrawal, PMP, Lauren Stoller, Manaswini Reddy, Nischita Prasad and Vijay Kumar K S for the collaboration, debates, debugging sessions and late-stage iterations that went into building QuickBrief. It was great experience going from "let's build an agent" to asking much harder questions: Does it work? Can we measure it? Can we explain it? Can we optimise it? And ultimately - can we trust it? Looking forward to taking these principles into the next AI system I build. #AgenticAI #GenerativeAI #AIEngineering #AIArchitecture #LLM #AIEngineering #Evals #AIOps #ContextEngineering #ArtificialIntelligence Course link - bit.ly/4j3FQ7p

Just completed an intense and incredibly practical journey into Agentic AI at LevelUp Labs. The biggest takeaway for me? Building an AI agent is the easy part. Building one that you can measure, optimise, explain and trust is the real challenge. For the capstone, we built QuickBrief - an AI analyst that takes a business question in natural language and turns it into an evidence-backed, decision-ready analysis. Not just text-to-SQL, but the ability to investigate why something happened through root-cause analysis. What made the experience particularly valuable was evolving the system through multiple levels of agency and seeing firsthand what each architectural decision really costs. A few learnings I'll take away from the course: 🔹 Performance first, optimisation second. Get the system correct and measurable before making it cheaper or faster. Otherwise, you're just optimising the wrong thing. 🔹 More sophistication doesn't necessarily mean more value. Autonomous or multi-agent architectures can look impressive, but the right amount of agency matters more than maximum agency. 🔹 Start with a workflow. A deterministic workflow is easier to understand, test and govern. It provides a reliable foundation before introducing autonomy where it actually adds value. 🔹 Context engineering matters. More information isn't necessarily better. Structuring the right context can be far more effective than giving an LLM access to everything. 🔹 Modularity makes Evals meaningful. Atomic, testable components make it possible to understand where something went wrong rather than simply knowing the final answer was wrong. 🔹 Observability changes how you build AI systems. Tracing isn't just about monitoring production. During our capstone, traces helped us identify real bottlenecks and failures, allowing us to optimise specific components rather than guessing. Perhaps my biggest takeaway: Don't build an agent because you can. Build the simplest system that reliably solves the problem and earn the right to add autonomy. A huge thank you to Aishwarya Naresh Reganti, Kiriti Badam, Ravi Yenduri and the entire team for making the course so grounded in real design principles and trade-offs. Special shout-out to my truly global capstone team Deepam Agarwal, Abhishek Agrawal, PMP, Lauren Stoller, Manaswini Reddy, Nischita Prasad and Vijay Kumar K S for the collaboration, debates, debugging sessions and late-stage iterations that went into building QuickBrief. It was great experience going from "let's build an agent" to asking much harder questions: Does it work? Can we measure it? Can we explain it? Can we optimise it? And ultimately - can we trust it? Looking forward to taking these principles into the next AI system I build. #AgenticAI #GenerativeAI #AIEngineering #AIArchitecture #LLM #AIEngineering #Evals #AIOps #ContextEngineering #ArtificialIntelligence Course link - bit.ly/4j3FQ7p

Sep 16, 2026

Sri Nagarajan

Kudos to Aishwarya Naresh Reganti and team for building a pragmatic and content rich AI training program. Your focus on fundamentals and resisting the urge to “sizzle for the sake of it” shows the Carnegie Mellon DNA come through! You guys have built a strong foundation and I hope you will continue to drive positive outcomes as new generations of decisions makers decide when, where, how much and why AI based solutions need to be deployed. There are a few more facets to consider adding to further strengthen what you are already doing, but that’s another day’s topic. Nice work LevelUp Labs!

Sep 16, 2026

Raju Sg

🚀 Just wrapped up the course lnkd.in/gJCVNghB , co-instructed by Aishwarya Naresh Reganti and Kiriti Badam! 💡 I highly recommend this program to engineering teams and leaders looking to design, deploy production-grade AI solutions. Rather than defaulting to standard paradigms, the curriculum drives deliberate architectural reasoning across every critical layer—from Context , prompt design, model selection, RAG, and MCP, to memory frameworks, guardrails, and autonomous agent orchestration. Key takeaways from the experience: Architectural Rigor over Hype: Encourages deep evaluation of trade-offs, observability, and evaluation mechanisms before building. Hands-On Practice: Direct experimentation with open-source tech stacks to solve real operational hurdles. Collaborative Community: Functions as an active peer network to share insights, discuss enterprise adoption, and filter through emerging trends together. Kudos to Aishwarya and Kiriti for delivering such a well-structured and impactful learning environment! #ArtificialIntelligence #AgenticAI #SystemDesign #MachineLearning #AIArchitecture #SoftwareEngineering Celebrating my new certification!

Sep 16, 2026

Siyamala Subramanian

Just wrapped up a 5-week Maven course on Building Agentic AI Applications with a Problem-First Approach — and came away with a much stronger understanding of what it really takes to move from an AI demo to a reliable agentic system. My biggest takeaway: building the agent is only one part of the problem. The real value comes from designing the right workflow around it — clear tool boundaries, evaluations, observability, monitoring, governance, and continuous optimization. For my capstone, I built Data Steward AI, an agentic data quality and governance application that: • Investigates customer data inconsistencies across systems • Analyzes breaking data-contract and schema changes • Gathers evidence using governed, read-only tools • Recommends the appropriate action • Keeps data changes behind explicit human approval • Maintains a complete audit trail The project gave me hands-on experience with: • LangGraph for agent orchestration and tool calling • Structured data, schema contracts, and API/tool integrations • Code-based and LLM-based evaluations • Arize/OpenTelemetry tracing and observability • Latency, token usage, tool-call, and agent behavior analysis • Guardrails around agent actions and data modifications • Iterating on prompts and workflows based on evaluation results rather than intuition One of the most valuable learnings was seeing how evals and observability provide much better control and visibility into agentic applications — where the agent is succeeding, where it is taking unnecessary steps, why a tool was called, and where the workflow can be improved. I’m leaving the course more confident in designing agentic AI systems with reliability, governance, and measurable outcomes built in from the start. 💻 GitHub: lnkd.in/gcQjSxad For anyone interested in the course, here’s the link: bit.ly/4yDXVO1 A big thank you to Maven team Aishwarya Naresh Reganti, Kiriti Badam for putting together such a practical and thoughtful learning experience. Thank you Ravi Yenduri, Sahana & entire maven team for your guidance. Excited to keep building at the intersection of Data Engineering + Agentic AI! #AgenticAI #AIEngineering #DataEngineering #DataGovernance #GenerativeAI #LLMOps

Sep 16, 2026

Divya Srinivasan

### Why I enrolled With Agentic AI, I wanted to understand **how to approach a problem, design the right architecture, evaluate the solution, and optimize its performance**. Tools like Claude Code can build an Agentic AI application in minutes, but I wanted to go beyond simply generating an application — I wanted to **understand how to build one from scratch and truly understand what’s happening under the hood**. ### A breakthrough moment The **aha moment** was realizing that instead of just learning concepts in isolation, the step-by-step approach to solving problems—along with understanding **when to apply each concept and how to use evals**—is a game changer at every stage of building agentic AI applications. ### What I’ll apply My biggest takeaway is the **problem-first approach** — starting with the problem, then designing the architecture using everything I’ve learned throughout the course. I now think about **performance first**, and then optimize for cost and latency. Finally, applying the right **evaluations (evals)** to validate the system. Most importantly, I now feel genuinely confident about **building Agentic AI applications end-to-end**. Aishwarya Naresh Reganti @Kiritibadam #maven #problemfirstapproach #agenticai

Sep 16, 2026

Gaurav Pahuja

Just wrapped up Building Agentic AI Applications with a Problem-First Approach with Aishwarya Naresh Reganti and Kiriti Badam What I really enjoyed was the focus on going beyond just building with AI and thinking through the problem, system design, and end-to-end lifecycle of agentic applications. Even while working closely with AI products, the course helped me step back, explore new perspectives, and understand parts of the lifecycle that we don't always get to own end-to-end in a corporate role. It’s a demanding course if you want to extract real value from it, but absolutely worth the time and effort. And believe me, with busy schedules throughout the year, there’s probably no perfect time to do it. Better to start now. It was even more rewarding to see our capstone team finish as the runner-up. I was happy to take on the lead role and work alongside Tanya Shukla, Gagandeep Pahwa, Barnali Patnaik, and Rohan Saxena to bring the project together. Thanks Kiriti and Aish for putting this together, and to the entire cohort for the learning and discussions. 🔗 lnkd.in/gwm_qRZp

Sep 16, 2026

Tanya Shukla

Enterprises want deterministic outcomes. AI gives you probabilistic ones. Bridging that gap is the real job! Two weeks into "Building Agentic AI Applications with a Problem-First Approach" with Aishwarya Naresh Reganti and Kiriti Badam and this framing really resonated with me. It's the clearest articulation of why every eval, calibration cycle, guardrail and trusted data pipeline is key in AI product design. These are all just attempts to close the gap between what the business needs and what the technology actually gives you. Thanks Aishwarya Naresh Reganti and Kiriti Badam for the great insights so far! #ProductManagement #AI #AgenticAI #EnterpriseAI #ContinuousLearning

Sep 16, 2026

Eric Spiegel

Thoroughly enjoyed the Maven Agentic AI builder course (link in comments) led by Aishwarya Naresh Reganti and Kiriti Badam. They made it a fun experience taking a "problem first" approach to topics such as Model Context Protocal (MCP) and AI Routers. The course was originally build for engineers using LangGraph and then they adapted it adding a track for non-coders using LangFlow. I found the some lecture content a bit deeper than I needed, so I just skimmed (lectures are recorded) and was able to learn the basics about building enterprise agentic AI workflows - enough for me to better advise computer science students on what to expect in their internships and first jobs. If you are a software engineer with little to no experience building enterprise applications with AI agents I highly recommend this course. Plus it's great networking! There were over a hundred students from across the globe.

Sep 16, 2026

Raj Barai

Since we started working with Aishwarya Naresh Reganti and Kiriti Badam on a problem-first approach to building agentic systems, one trade-off has kept surfacing: control versus autonomy. The more decisions you leave to an agent, the more flexibility you gain, but the harder its behaviour can be to predict. I have now built two AI systems in very different ways, and the result ran opposite to what I expected. The first was a RevOps system for my team, vibe-coded over three weeks in the gaps around a full-time job. One Google Sheet as the database, 2,400 lines of Apps Script and one HTML file. It was deterministic by design. I assumed that meant I controlled it. But I had built it by describing what I wanted, through text and voice, and letting an LLM write the code. Every fix and change went back through the same process. The software behaved predictably. My ability to understand and change it did not. The second system, built with the team at LevelUp Labs was deliberately more autonomous: part deep-research agent, part RevOps assistant. Yet I had more control over it. Not because autonomy creates control, but because the system was: 1. Decomposed into visible components 2. Configurable at the component level 3. Observable through traces 4. Easier to diagnose and modify I could see the router and its named routes. I could set the retriever’s chunking strategy. I could choose a different LLM for each step based on cost, latency and accuracy. I could define tool inputs and trace where a failure began. You cannot evaluate what you cannot observe. And you cannot make reliable what you cannot evaluate. If I rebuilt the RevOps system now, I would first scope the capability and identify where judgment was actually needed. Then I would decide which steps belonged in deterministic code and which genuinely required an agent. When building AI systems, autonomy is not the measure of sophistication. Reliability is. Week 4 of Building Agentic AI Applications with a Problem-First Approach. #AgenticAI #AppliedAI #AIStrategy #RevOps #BuildInPublic

Sep 16, 2026

Pankaj Thakur

@thakur2720

Thanks to a great AI course by @aish_reganti and @kiritibadam. The AI chaos is exhausting — new tools, new hype, new FOMO, every single day. This course did something different: it pulled me back to basics. Sometimes the answer isn't more noise. It's clarity.

Sep 12, 2026

Pratik

@PratikLondon

Thank you @aish_reganti and @kiritibadam for delivering a wonderful course - maven.com/aishwarya-kiriti/genai-system-design . Loved all the aspects- Problem focus, Foundations setup, architecture levers, framework considerations, community support, project collaboration and your enthusiasm, efforts

Sep 11, 2026

Abhi Gamer

@gameabhi

Just wrapped up 5 weeks of "Problem First Approach to Building AI Applications" with Aishwarya Reganti (@aish_reganti) & Kiriti Badam (@kiritibadam), and it's easily one of the most practical AI courses I've done. (I've been sharing my key takeaways each week, if you want the fuller picture.) The breadth alone was substantial (RAG, evals, agent arch, memory, MCP, all covered); what set it apart was the problem-first approach running through all of it. Every framework, every trade-off, every design pattern was taught by starting with "what problem does this actually solve" rather than "here's a tool, go use it." That's rare, and it's exactly what you need when the field moves this fast. Content stayed genuinely current throughout, right down to references to protocol changes from just months ago, and the guest lectures (our batch got lucky with AI Summit happening at the same time) added real practitioner perspective on top of the core curriculum. Whether you're a tech leader trying to cut through the AI hype or a builder wanting to actually design and ship these systems, this course delivers both the mental models and the hands-on depth to back it up. ⭐️⭐️⭐️⭐️⭐️ maven.com/aishwarya-kiriti/genai-system-design

Sep 8, 2026

andi schuster

@andischuster

AI systems design with @aish_reganti @kiritibadam is an absolute goldmine of relevant skills, knowledge and frameworks for AI PMs. Everything is based on first principles combined with relevant case studies form real practicioners. Highly recommend. maven.com/aishwarya-kiriti/genai-system-design

Sep 7, 2026

Vidya P.

Last month I wrapped up the "Building Agentic AI Applications with a Problem-First Approach" course on Maven, and it exceeded all expectations! 🚀 What sets this program apart is the focus on real-world constraints—learning how to balance model performance, latency, and costs rather than just chasing the newest tools. The focus on context engineering and robust Evals gave me actionable frameworks I can apply immediately. Huge thanks to the incredible instructors, Aishwarya Naresh Reganti and Kiriti Badam for sharing their deep enterprise experience and keeping the sessions so engaging. Highly recommend this to any engineer, PM, or leader looking to move past the AI hype and build production-ready agentic systems! 🔗 Course link: lnkd.in/e2veuk86

Jul 17, 2026

Tharun Mothukuri

The past five weeks were packed with some of the most hands-on AI learning I've done. Excited to share that I completed "Building Agentic AI Applications with a Problem-First Approach" on Maven. Here are my top 3 takeaways: 1/ Start with the business problem, not the technology. The biggest shift for me was learning to evaluate tradeoffs between model choices, cost, and latency before writing a single line of code. 2/ Evals are the backbone of reliable AI systems. Building a framework to measure failure modes early saves so much guesswork and makes iterative improvement practical. 3/ Guardrails matter more than I initially thought. Setting boundaries around model behavior, handling edge cases, and preventing misuse are things you really have to plan for upfront, not as an afterthought. Coming from an infrastructure background, I never thought I'd be building AI applications. This course connected the dots between the data platforms I manage every day and where AI is heading. The line between the two is getting thinner every day. Here's the part I'm especially proud of: Our capstone project *Natural Language to SQL* finished as the runner-up! Huge kudos to all my team members (Sridhar M.J, Vidya P.,Cyndi (Xinyi) Zhang and Jo Mig). It was a great experience building something real together and seeing it recognized among so many talented teams. Thank you Aishwarya Naresh Reganti and Kiriti Badam for building a curriculum that is genuinely hands-on, practical and grounded in real enterprise experience. #GenAI #AgenticAI #AI #MachineLearning #ContinuousLearning

Jun 11, 2026

Vijay Majeti

I recently completed "Building Agentic AI Applications with a Problem-First Approach" on Maven by Aishwarya Naresh Reganti and Kiriti Badam. If you're working in agentic AI — or planning to — this course is worth your time. What stood out: they don't start with tools or frameworks. They start with the problem. Define your metrics. Understand what "good" looks like before you write a single line of code. That mindset shift alone made this valuable for me. Our cohort's capstone was an 𝗔𝗜-𝗣𝗼𝘄𝗲𝗿𝗲𝗱 𝗣𝗿𝗲-𝗧𝗿𝗮𝗱𝗲 𝗦𝗶𝗴𝗻𝗮𝗹 𝗕𝗿𝗶𝗲𝗳𝗶𝗻𝗴 𝗦𝘆𝘀𝘁𝗲𝗺 — correlating live market news to tickers to help traders act on the right information faster. As a team we took it from a raw idea through three meaningful iterations — a great demonstration of the course's iterative, eval-driven approach in practice. I work in agentic AI, and I still walked away with sharper thinking on guardrails, grounding, and how to evaluate systems that actually matter in production. Thank you Aishwarya Naresh Reganti and Kiriti Badam — this course genuinely changed how I think about building agentic systems from a problem-first approach. 📚 Course: lnkd.in/geii7hTx 🎞️ Slide Deck: lnkd.in/gmpimGkn 💻 Code: lnkd.in/gRCejW5j #AgenticAI #LangGraph #ProblemFirstAI #MavenCourse #AIEngineering

Jun 8, 2026

Veronika Benkeser

Building agentic systems with our engineering team and recently completing the "Building Agentic AI Applications with a Problem-First Approach" Cohort on Maven has sharpened how I think about durable AI capabilities in production: ➡️ Start with the business problem. ➡️ Define success criteria. ➡️ Build the simplest implementation that solves the workflow. ➡️ Define failure modes, constraints, and escalation paths. ➡️ Build evals. ➡️ Design guardrails around inputs, outputs, tool use, permissions, and escalation. ➡️ Trace execution. ➡️ Evaluate intermediate steps, not just final outputs. ➡️ Add autonomy only where it earns its complexity. ➡️ Expand edge cases. ➡️ Ship with human review, versioned changes, fallback paths, and rollback ability. ➡️ Monitor and calibrate based on production traces: failures become evals, evals guide architecture and guardrail changes, and each iteration strengthens the system boundaries so issues in planning, retrieval, tool use, or handoffs are detected and contained before they create larger user-facing failures. What I appreciated most about this cohort experience was its focus on how agentic systems actually fail in production. It was fun to get hands-on, especially debugging my Agentic RAG implementation - it turned out to be a retrieval issue, just like most RAG issues! My team’s capstone project, the Agentic Change Impact Analyzer, brought the learning together: 3 architecture iterations, tools, evals, guardrails, observability, user trust, and feedback loops as one system designed to reason across large codebases and assess how a proposed feature or a code/infrastructure/policy change would affect different parts of an application. A sincere thank you to Aishwarya Naresh Reganti and Kiriti Badam for bringing real technical depth to every session and pushing us to think deeper and build bolder. Thank you to the Chai & AI community, guest lecturers, and my capstone team - Shivateja Madipalli, Shishir Kumar, Anshul Kansal, Deepika Srinivasan, Ramakrishna Vedantham, Shilpa Sharma - for your insight, debate, and the incredible shared energy! I highly recommend this program: lnkd.in/gkvt-Qgd #GenerativeAI #AgenticAI #AISystemDesign #ArtificialIntelligence #TechnologyLeadership #ContinuousLearning

Jun 8, 2026

Jorge Alvarez

Tools will change overnight, but a rigorous thought process and a dedicated community are what endure. Just wrapped up an incredible 5 weeks with Building Agentic AI Applications with a Problem-First Approach. Having taken AI and agent-native programs across MIT, UT, edX, and Kaggle, this has easily been my favorite course to date. The level of intent behind the frameworks, labs, and tooling—paired with a remarkably open, collegial culture—blew me away. If you’re looking to cut through the noise and focus on real AI fundamentals, look no further than Kiriti Badam and Aishwarya Naresh Reganti. Brilliant execution. 🚀 Check out the course here: lnkd.in/gxwxUH5T

Jun 8, 2026

Subbu Venkataraman

I enjoy learning from experienced practitioners who have worked hands-on in building agentic AI frameworks. I’m happy to share that I completed the Building Agentic AI Applications with a Problem-First Approach course led by Aishwarya Naresh Reganti and Kiriti Badam. Their in-depth knowledge and practical experience in this subject made the course very valuable. Some of the things I built and practiced include: · Creating a workflow agent in Langflow that can classify user inputs and route them appropriately · Leveraging conversation memory to provide contextual responses · Creating an autonomous agent that uses web search or RAG based on user questions · Using a multi-agent architecture to create an AI assistant capable of doing deep research · Getting hands-on exposure to OpenClaw, Composio, and the Telegram tool Thank you Aish and Kiriti for sharing your knowledge and passion throughout the course. #AgenticAI #GenerativeAI #AIStrategy #ContinuousLearning

Jun 8, 2026

Luca Zorzenon

I just finished "Building Agentic AI Applications with a Problem-First Approach" and what stuck with me was how much of building AI systems is evaluation, and how it does not stop once the thing is live. I expected testing an AI system to work like testing software: get a version running, confirm it does what it should, move on. It does not. Because the output is non-deterministic, one round of testing tells you very little. You evaluate continuously: define what "good" means, compare each version against the last, and keep recalibrating as you learn how the system actually behaves. That is a product question as much as an engineering one. Reliability, latency, cost, and human review are not implementation details you sort out later: they decide what is worth building in the first place. The capstone project made it concrete. My team built an AI-powered pre-trade briefing system for market news, and I worked on the implementation for the live demo. The real problems were not only in the code: briefings drifting from the source, the same market signal surfacing more than once, and indirect events that were hard to call as relevant or noise. Each was a failure mode we had to define, measure, and design against before the output could be trusted. The habit I am keeping is to work out what can break, and how I would catch it, before reaching for more complexity. On our project, that made me much more wary of adding agentic layers before the basic workflow was observable and testable. A real thanks to Aishwarya Naresh Reganti and Kiriti Badam. This space moves fast and it is easy to drown in it. The way they structured the content, the lessons, and the hands-on work left me with a mental map that holds up as the field keeps moving. Course link in the first comment.

Jun 8, 2026

Seena Iype

AI for Day Traders - this was our capstone project. I recently took “Building Agentic AI Applications with a Problem-First Approach” on Maven by Aishwarya Naresh Reganti and Kiriti Badam. The course teaches how to approach AI with a problem first approach and use AI in a more iterative manner making sure you are clear on metrics for evaluations are in place for building and scaling. ----- My team built an 𝗔𝗜-𝗣𝗼𝘄𝗲𝗿𝗲𝗱 𝗣𝗿𝗲-𝗧𝗿𝗮𝗱𝗲 𝗦𝗶𝗴𝗻𝗮𝗹 𝗕𝗿𝗶𝗲𝗳𝗶𝗻𝗴 𝗦𝘆𝘀𝘁𝗲𝗺 - a Day Trading add-on that provides traders indicators from relevant news sources and correlating them to tickers, helping them to take quick decisions and tap onto opportunities. We did this across three iterations: → Iteration 1 — Workflow Agent: fetch news, extract events with an LLM, cluster by ticker, generate grounded briefings, validate with an LLM judge → Iteration 2 — Catalyst Memory: semantic deduplication via a Catalyst Memory Ledger — same story doesn’t brief twice → Iteration 3 — Exposure Graph: detect indirect catalysts from geopolitical, supply chain, and macro events — only routed to a ticker when a graph path supports it --- It was amazing to see how a small group of engineers and product folks met across 4 timezones every day and got the prototype ready. It was also so good to see how we also talked about security, compliance as part of this project. Special credit to Luca Zorzenon and Ankita Mehta for building the codebase 📚 Course: lnkd.in/geii7hTx 🎞️ Slide Deck: lnkd.in/gmpimGkn 💻 Code: lnkd.in/gRCejW5j 🌟 Golden Dataset: lnkd.in/gPACrJA3 Wonderful working with you all - Vijay Majeti, Roshina Mohamed Rafee, Sita and Ravi. Lets continue the learning and discussions, as the AI world is changing and its changing the tech landscape :)

Jun 8, 2026

Vinay Hemmady

Just wrapped up Building Agentic AI Applications with a Problem-First Approach taught by Aishwarya Naresh Reganti and Kiriti Badam through Maven. Over the past few years, AI has evolved at an incredible pace. Amid all the excitement, one of my biggest takeaways from this course was that successful AI systems are built on strong engineering and architectural fundamentals, not just the latest models, frameworks, or tools. Three lessons stood out for me: ✅ Start with the problem, not the technology. The best solutions begin with understanding the business need and designing the simplest workflow that delivers value. ✅ System design matters more than ever. As AI accelerates implementation, architectural thinking, trade-off analysis, and sound engineering judgment become even more important. ✅ Evaluation is critical. Building AI applications isn't just about generating outputs; it's about creating reliable, safe, and measurable systems that can operate at scale. What I appreciated most was the course's practical, hands-on approach and the opportunity to learn alongside an exceptional group of professionals. A sincere thank you to Aishwarya Naresh Reganti and Kiriti Badam for creating such a thoughtful and impactful program. Special thanks to Ravi Yenduri and Sahana Venkatesh for their outstanding support throughout the journey. I look forward to applying these learnings as we continue exploring how AI can create meaningful business value while maintaining the engineering rigor required for enterprise adoption. For anyone looking to move beyond the hype and build a strong foundation in applied AI systems, I highly recommend this program. lnkd.in/g9XFWfNC #GenerativeAI #AgenticAI #AISystemDesign #EngineeringLeadership #ArtificialIntelligence #TechnologyLeadership #ContinuousLearning

Jun 5, 2026

Ashwin Dinoriya

Solid Saturday ROI. Finished the Agent-Native Operator Bootcamp (Aishwarya Naresh Reganti & Kiriti Badam on Maven ). Worth the time. I’d used #OpenClaw intermittently for learning, but always wondered if my setup was actually locked down. This course killed that doubt—step-by-step walkthrough covering the entire deployment from security-first principles. No handwaving. Six hours of content. Felt like a week of real engineering work compressed. Built a production-ready “Partner Agent” config I’m actually confident deploying. Tldr: If you’re running #OpenClaw in production or planning to, take this. Cuts out the guesswork and the second-guessing. #GenAI #AgenticAI #OpenClaw #AIEngineering

Jun 5, 2026

Vaibhav Gupta

Excited to share that I completed the Building Agentic AI Applications with a Problem-First Approach course — and even more grateful that our team won the capstone for our project, Customer Escalation Resolution Copilot. 🏆 I joined this course to get more hands-on with AI and better understand how agentic systems can be applied to real product and workflow problems. What stood out most to me was the emphasis on starting with the problem first — understanding the user, the workflow, the decision points, and the risks before jumping into agents, RAG, or automation. For our capstone, we focused on a practical support workflow: helping tier-1 support engineers resolve escalated customer tickets faster and with better context. Today, engineers often have to manually piece together information across past tickets, docs, chat threads, and logs before they can even begin diagnosing the issue. Our idea was to create an evidence-grounded copilot that could retrieve relevant context, draft evidence-backed recommendations, surface gaps, and keep the engineer in the loop for the final decision. This project helped me better appreciate that building useful AI products is not just about model capability. It is also about workflow design, trusted data, evaluation, guardrails, latency, feedback loops, and knowing where human judgment still matters. I especially enjoyed contributing from a product and workflow perspective — thinking through the escalation process, how the experience should fit into an engineer’s day-to-day workflow, and how feedback could make the system more useful over time. Huge thanks to my teammates Abhishek Raj, Evelyn Chou, VENKAT SUMANTH REDDY BOMMIREDDY, Akansha G., @Vidya S, and Vidyasagar Sirigireddy for the thoughtful collaboration, strong execution, and shared learning throughout the capstone. And a big thank you to Aishwarya Naresh Reganti, Kiriti Badam, and Ravi Yenduri for the guidance throughout the course, the hands-on assignments, and for creating such an engaged Chai & AI learning community. Course link: lnkd.in/gD8JqBim Looking forward to continuing to build on this foundation and applying these concepts more deeply across AI products, data platforms, and workflow automation use cases. #AgenticAI #GenerativeAI #AIProductManagement #RAG #AIAgents #WorkflowAutomation

Jun 5, 2026

Vidhya Rajan

We presented our AI capstone today — and I couldn't be more proud of what this team pulled off. Not because of the product we built, though I'm proud of that too. But because of the team Jorge Garcia, Rohit Iyer, Kawal Grower, Priyank Patwa and Gokul Panicken that made this happen despite of timezone and availability challenges. What came through was our collective purpose of leveraging AI to solve real problems beyond POC's, we all come from completely different worlds — different industries, different skill sets, different mental models of what "building with AI" even meant. That diversity became our biggest advantage, challenging ourselves to think about the problem under the lens of the varied roles and domains. To Aishwarya Naresh Reganti, Kiriti Badam and Ravi Yenduri— you didn't just introduce us to Building Agentic Solutions with a Problem First approach - you built a space where confusion was expected and curiosity was rewarded. That's rare. Thank you. This is what learning should feel like. More to come on what we built, what broke, and how it has helped and how I plan to leverage it professionally. Stay tuned. #GenAI #AIatScale #Maven #EnterpriseAI

Jun 5, 2026

Evelyn Chou

P0 ticket escalated twice before your first coffee. Sound familiar? If you're a PM, Engineer, or Support specialist, you know the drill. The questions come fast: Is there a repro? System-wide impact? How much ARR are we risking if resolution takes 30 minutes vs. 3 hours? Here's the problem: Everyone's building AI support agents. Fewer are asking whether a multi-agent system is actually the right solution, or if it's just expensive complexity pretending to be innovation. This weekend, my team (Vaibhav Gupta, Abhishek Raj, Akansha G., Venkat, and Vidyasagar) is presenting our capstone project: a problem-first approach to customer escalations that keeps support engineers in control instead of buried under AI orchestration overhead. We'll show: 👓 How we went from deterministic to creative iterations 👓 What "problem-first AI" means in production support context The demo is open to all: PMs, engineers, support leaders, or anyone who's tired of "AI will solve it" as a substitute for systems thinking. See you this Saturday 5/30. Link in comments. Huge shoutout to the course team Aishwarya Naresh Reganti, Kiriti Badam, Ravi Yenduri #EnterpriseAI #AIAgents #MultiAgentSystems #Workflow #RAG

Jun 5, 2026

Kapil Gupta

As a software engineer, Generative AI felt like a massive paradigm shift — and finding structured guidance through all the noise was genuinely hard. That changed when I completed "Building Agentic AI Applications with a Problem-First Approach" What would have taken me 3–4 months of piecing together free resources was compressed into a few focused weeks — with the right intuition built along the way. What stood out most was the problem-first approach. This course didn't just teach tools — it taught me how to think about AI systems in the middle of all the hype around endless AI tools and services. I can now navigate new developments without panicking, cut through the marketing hype, and participate in AI architecture discussions with real confidence. For engineers and tech leaders responsible for building and owning technical roadmaps, that's an invaluable shift. The production use cases from guest lectures added grounding that's genuinely hard to find anywhere else. The capstone — AI-powered Customer Support and Operations Agent — brought everything together. Designing, debugging, and optimizing an end-to-end agentic pipeline from first principles was the best way to consolidate the learning. A sincere thank you to Aishwarya Naresh Reganti (lnkd.in/g7AgT8bB) and Kiriti Badam (lnkd.in/gqiUp9fj) for the depth and care you brought to every session. And to the Chai & AI community — the discussions and support made this journey far more meaningful than any solo learning path could have. If you're an engineer trying to cut through the AI noise and build real intuition, this course is worth your time: lnkd.in/g3xR4b_a #AgenticAI #AIEngineering #GenerativeAI #ChaiAndAI #SoftwareEngineering #LearningJourney #ArtificialIntelligence #NLP #CapstoneProject

Jun 5, 2026

Minh Vu

Minh Vu, CFA, CFP

Just completed the “Building Agentic AI Applications with a Problem-First Approach” course by Kiriti Badam and Aishwarya Naresh Reganti The biggest value of this course for me is the framework. Agentic AI can feel confusing because there are so many tools, models, and architectures. The landscape is changing so quickly that the tool used today might become outdated tomorrow. What this course provides is not just about tool-based approach, but a clear structure for thinking about how to build AI solutions. Starting from the problem, understanding the workflow to designing the AI system around business needs. Those thinking principles are what will stay valuable even as the technology changes. Recommended for anyone building, or wanting to learn how to build agentic AI applications.

Jun 5, 2026

Sankar Kumar Palaniappan

🚀 Just wrapped up "Building Agentic AI Applications with a Problem-First Approach" on Maven(lnkd.in/ehXpD4qV) and it's easily one of the most impactful learning experiences I've had this year. In a space crowded with surface-level GenAI content, this cohort stood out by teaching the why and the how, not just the what. The course is built around a decision-map framework drawn from the instructors' experience shipping 50+ AI products which means every concept is grounded in real enterprise tradeoffs, not theoretical checklists. What the course covered & what I walked away with: 1. Problem-first AI intuition for enterprise use cases knowing when agentic AI actually adds value (and when it doesn't) 2. Evals-driven mindset framing AI problems through measurable outcomes rather than features 3. Practical prompt and context engineering decomposition, meta-prompts, LLM judges, and semantic scoring 4. Building and optimizing agentic pipelines retrieval, memory, multi-agent coordination, and protocols like MCP and A2A 5. A real-world capstone designing an end-to-end agentic search system across three iterations, integrating RAG, MCP, and multi-agent components. ❤️ Huge thanks to Kiriti Badam and Aishwarya Naresh Reganti— Kiriti's deep technical mastery brought rare clarity to the systems-thinking side of the course, making complex architectural tradeoffs feel intuitive. Aish's problem-first framing reflects an exceptional command of the enterprise AI landscape, turning abstract concepts into a sharp, decision-oriented playbook. The coordination across the cohort live sessions, office hours across time zones, two assignment tracks (code and low-code), and a thoughtfully structured capstone was outstanding. The flipped classroom format with 35+ hours of live interaction made it feel less like a course and more like an apprenticeship. #AgenticAI #GenerativeAI #AIEngineering #Maven #ContinuousLearning #LLM #MCP #AIProductManagement

Jun 5, 2026

Kriti Bains

Just completed the Maven course by Aishwarya Naresh Reganti and Kiriti Badam. Instead of feeling overwhelmed, I came out of it feeling more confident to experiment, build, and think ‘Problem-first’! If AI content has ever felt overwhelming or you’ve wanted to build something but didn’t know where to start, this course is worth checking out. lnkd.in/gpNGiZkV Thank you Aish and Kiriti for creating such a grounded and accessible learning experience. #AI #AgenticAI #Maven #AIBuilders

Jun 5, 2026

Muttu Shivayogimath

I attended this course, and learnt a great deal.. I can try to build my Agentic AI app, on my own, without taking LLMs help Thank you team lnkd.in/ghNbBU5v Kiriti Badam & Aishwarya Naresh Reganti Building Agentic AI Applications with a Problem-First Approach by Aishwarya Naresh Reganti and Kiriti Badam on Maven maven.com

Jun 5, 2026

Tarah Srethwatanakul

Just started Aishwarya & Kiriti's Maven course on building agentic AI applications, and I love the way they've framed the content: Models, agents, harnesses are treated as tools to solve the problem, rather than the thing to learn. The focus is on the problem / use case, and a recurring theme is "if we choose this approach, what's the trade-off?" There's a capstone where I'll be building an agentic AI app for an enterprise use case — the demo is public, watch this space! 👀

Jun 5, 2026

Paola Maisonet Caballero

I recently earned this certificate and Ive learned so much in such a short period of time. First off, I want to thank Coral Cotto Negrón, PhD MS for believing so much in me and giving me a path to move forward in tech. She's one of those rare mentors life/God gives you to keep you accountable, while also teaching you how to get back up from your stumbles. She provided me the opportunity (and the privilege) to learn many many things, but also this course. Aishwarya Naresh Reganti Kiriti Badam Thank you for building such an amazing course and ecosystem filled with people who come together from all different backgrounds and cultures to learn how to build agentic solutions from a Problem First Approach - something that I struggled with and something that Coral knows Ive been working hard to refine. Ill be sure to take everything learned to my career at Cibergente and life all on its own.

Jun 5, 2026

Javed Jafar

@javedmj

Last 4 weeks have been an incredible learning journey with @aish_reganti & @kiritibadam , additional instructors ( Ravi , Sahana) and Course Ops ( Theresa & team) in their GENAI system design for becoming problem first builder. Highly recommend maven.com/aishwarya-kiriti/genai-system-design

May 23, 2026

Kyle M. Brown

I solve problems using technology.

I just completed "Building Agentic AI Applications with a Problem-First Approach" by Aishwarya Naresh Reganti and Kiriti Badam at Maven. The biggest takeaway wasn't a framework or a tool. It was this: ➡️ Intelligent software development doesn't replace what you already know about building software. It extends it. ➡️ The same architectural patterns — decomposition, evaluation, system design, managing tradeoffs — all apply. What changes is that your system now reasons instead of just executing. ➡️ Outputs are inferred, not hardcoded. Evaluation becomes your primary feedback loop. ➡️ Failure modes become design constraints, not bugs. Learning these fundamentals didn't feel like starting over. It felt like unlocking a new layer of patterns I've spent 20 years building intuition for. #ArtificialIntelligence #AgenticAI #SoftwareEngineering #AISystemDesign lnkd.in/epfzk9XH

Mar 12, 2026

Sjur Sundin

Product Leader shipping enterprise AI products that solve real problems

I just finished this class: lnkd.in/e9QZ8Ca7, and I highly recommend it. It’s challenging and will push you to get hands-on with modern AI tools, but you'll walk away with a much stronger understanding of how to think about AI and apply it to real business problems. What I especially liked was the focus on starting with the problem first, not the technology. The instructors were super engaged throughout, and the additional lectures with industry professionals helped us understand how AI is changing the roles and responsibilites in the workplace. Also, there is hands-on homework, and a big capstone project, and this made the experience so much more fun, you really have to stay focused throughout. Aishwarya Naresh Reganti Kiriti Badam

Mar 12, 2026

NS

Narendar Reddy Soma

Senior Manager - Data & AI @ Accenture

The "How" is getting easier, but the "Why" is everything. I spent 4 hours today immersed in the Capstone Project Demos as part of “Building Agentic AI Applications with a Problem-First Approach” cohort. After seeing 30+ product ideas brought to life, one thing is crystal clear: The idea is the moat. We have the tools and the tech to build faster than ever, but the real winners are those who approach AI with a problem-first mindset. Seeing these agents solve specific, high-friction enterprise challenges moving from problem scoping to iterative design reiterated that we aren't just building bots; we’re engineering reliable systems. Huge kudos to all the presenters in this cohort! The energy and execution were top-tier. A massive thank you to Aishwarya Naresh Reganti and Kiriti Badam for leading this mission and pushing us to build with purpose. lnkd.in/gi5DVHzz lnkd.in/ga2gVX3m #AgenticAI #GenerativeAI #ProblemFirst #LevelUpLabs #Maven

Mar 12, 2026

Daya Reddy

Success Management | Enterprise Tech | DoD | CoderDojo Champion

For the last 5 weeks (nights + weekends) I've been immersed in Building Agentic AI Applications with a Problem-First Approach course. What worked for me in this course: 1. Flipped Classroom Cohort with Office Hours In my busy professional life, I've struggled last year to prioritize AI learning, signing up for online courses (like deeplearning.ai) but not getting to the finish line. The flipped-classroom, cohort-based format created the accountability I needed for learning progress, along with collaboration with engaged mentors and peers. One note for busy professionals: the content is dense and the pace is intense. Watching videos or reading content at my own pace rather than sitting through lectures allowed me to engage more deeply with mentors and the community during office hours and Slack. 2. Problem is the Hero Not AI AI itself doesn't create value solving the right customer problems always does. This is not a course that teaches you how to build an AI agent in 30 minutes or chase the latest agentic tools like ClawdBot, OpenClaw, or the next new . The material is scaffolded from foundational AI concepts through context engineering to building agentic systems with MCP. You can think of it as meta-learning on AI with a Problem-First approach (the why), and for me reinforcing the Value-First (so what) lens, especially when 2025 reports showed 95% Gen-AI use cases failed to showcase value. 3. Capstone Project It was great collaborating with Mahendra Nimishakavi to bring it all together for a working prototype but focused on showcasing why, so-what and design iterations for demo day. They told us that “Pain is the new moat.” and it's true without reps you don't build the AI knowledge muscle. This gave me (and honestly forced me) the opportunity to step outside my work comfort bubble, carve time for learning (never stops) and build a problem-first AI design mindset to carry forward with the customers I support and into a PSSM practice we're scaling. Congratulations Aishwarya Naresh Reganti and Kiriti Badam for creating such an impactful course and fostering a great Problem-First AI (human) community. Course Link: lnkd.in/eAkCjR9s #ProblemFirstAI

Mar 12, 2026

Mahendra Nimishakavi

Senior Mobile Engineer at Shopify

🚀🚀 Learning a ton about Agentic AI systems from the course Building Agentic AI Applications with a Problem-First Approach A few sharp takeaways so far: • Non-determinism → AI systems aren’t predictable pipelines; they require guardrails, feedback loops, and probabilistic thinking. • Evals → if we can't measure we can't improve the system - evals are the new unit tests. • Problem-first design → Start with the user problem and not throw an LLM at everything and expecting it to do magic it’s definitely a new system design mindset and I am looking forward to apply these concepts in real world problems. #AgenticAI Kiriti Badam Aishwarya Naresh Reganti LevelUp Labs

Mar 12, 2026

Sathya Chackravarthy

Enterprise Data Strategy for AI

I recently attended the Problem First AI – GenAI System Design course, and I wanted to share a note of appreciation for the incredible thought and structure that has gone into designing this program by @Aishwarya Naresh Reganti and Kiriti Badam What stood out to me is how thoughtfully the course has been structured to cater to a wide range of learners — from people new to the field to experienced professionals and senior leaders. Striking that balance is not easy, but this program does it very well. I also appreciate the fact that the sessions are recorded and available for replay. In technology — and especially in AI — continuous learning is essential to stay ahead of the curve. This course does a great job of building a solid foundation with a true problem-first approach to AI system design. Highly recommend this course to anyone serious about understanding and applying Generative AI in a structured way. Course link: lnkd.in/eahVveuk #AI #GenerativeAI #AISystemDesign #ContinuousLearning #MachineLearning

Mar 12, 2026

Mirna Ortiz Treviño

Product Manager | Ex-Uber

I just finalized the GenAI System Design course, and it has been a game-changer for how I think about building agentic applications. During the program, I developed Parrot, a reminder app built to streamline household chores. By leveraging Langflow and Arize as my agentic backbone, I was able to architect a system that doesn't just follow instructions but truly manages the process of task management for a household. A massive thank you to Aishwarya Naresh Reganti and Kiriti Badam for providing the framework to build truly robust AI systems. The focus on flywheel-inducing evals is exactly what’s needed for anyone serious about moving AI projects into production. I’ll be sharing more about my technical learnings and the development process soon (once I officially overcome my "LinkedIn cringe" 😅). Check it out here: lnkd.in/gAwp_Xyp #AI #MachineLearning #SystemDesign #Innovation #Maven #AgenticAI

Mar 12, 2026

Anna Mogiłło-Dettwiler

Pilot | Flight Dispatcher | Computational Linguist

✈️ Six years ago, my pilot training was suspended because of COVID. What started as an unexpected pause turned into a brand new chapter: I wrote my first lines in Python. 🐍 Later, studying Computational Linguistics at University of Zurich felt like connecting the dots between my master’s in Applied Linguistics and the world of programming. Back then, every small coding task meant hours of searching online for solutions to simple problems (why is this for-loop not working again?). 🏄🏻‍♀️Fast forward a few years and I find myself coding again, this time in the middle of the GenAI wave. 🎉 Here’s how I got back on board: I’ve just completed the Problem-First AI course by Aishwarya Naresh Reganti and Kiriti Badam lnkd.in/eKz4CC2j 📖 It was a great deep dive into designing real agentic systems: - starting with the problem first - thinking about evaluation and reliability - understanding when agents actually make sense. 💡One unexpected shift: I started using Cursor. Writing code is now incredibly fast, sometimes faster than your own thoughts. Which makes one thing clear: staying disciplined and keeping the human in the loop matters more than ever (especially in aviation). Exciting times to start building. Oh, and if you’d like to see what we’ve built, there’s an open Capstone Project session this Saturday, 6:00 pm Zurich time. Our team built the Autonomous Household Chief of Staff agent - sounds mysterious? 🧐Check it out, then! lnkd.in/e-nxDDGF

Mar 12, 2026

Sourabh Sanghi

AWS Certified Developer , Architect | DEVOPS | JAVA ENGINEER

Steve Jobs famously said: “You’ve got to start with the customer experience and work backwards to the technology.” In the current GenAI hype cycle, it’s easy to do the opposite—falling in love with a new model before fully understanding the problem it’s meant to solve. I’m currently spending my time challenging that "technology-first" trap while taking the course "Building Agentic AI Applications with a Problem-First Approach" with LevelUp Labs The program is a refreshing deep dive into the discipline required to build AI agents that actually work in production. Instead of just chasing "wow" moments, we are focusing on building robust system designs that prioritize reliability and business value. Here are the core frameworks that are changing my approach: The Agency-Control Trade-off: I am learning that higher autonomy is a privilege earned through reliability. We must carefully balance an agent’s power to act with the level of control we can realistically maintain to ensure system trust and safety. Model Evals vs. Product Evals: While model benchmarks are useful, Product Evals are the essential component for building a reliable agentic system. I am learning how to implement an Evaluation Flywheel to iteratively refine performance, ensuring the AI actually meets specific product requirements rather than just performing well in a vacuum. A huge thank you to Aishwarya Naresh Reganti and Kiriti Badam for sharing such practical, enterprise-ready insights. Their framework is helping me move past the "demo" phase to understand how to build truly scalable, autonomous systems. For my colleagues and friends in the AI space looking to sharpen their system design skills, I highly recommend checking this course out. Find the course and upcoming cohorts here: lnkd.in/gxTedvHa #GenerativeAI #AIAgents #SystemDesign #LevelUpLabs #ProblemFirst #AIEngineering #AIProductEvals

Mar 12, 2026

Sricharan Sunkara

MBA | Product Manager, AI | Hands-On Technical Leader

I just wrapped up an incredibly intense and rewarding 5-week bootcamp: Building Agentic AI Applications with a Problem-First Approach on Maven, taught by Aishwarya Naresh Reganti and Kiriti Badam. I went into this 5-week program with high expectations, and it completely exceeded them. This wasn't just a typical course; it felt like entering a new world of problem-solving. Instead of getting distracted by shiny new tech, we learned the thinking behind the tools. By starting small and layering on higher-level concepts, I learned how to effectively act as the architect while letting AI do the heavy lifting. A massive shoutout to the amazing community, the insightful guest lecturers, and my capstone project team. The formal course might be over, but the learning and Slack conversations are definitely continuing. I’m looking forward to applying these AI frameworks to process improvements and system architecture moving forward. Check out the course here : lnkd.in/gADj48-P Follow them on LinkedIn: Aishwarya Naresh Reganti | Kiriti Badam X: @aish_reganti | @kiritibadam

Mar 12, 2026

Ayush Kumar

Product & Growth Leader - Consumer Tech & PLG SaaS | Human psychology | Reforge alum | Curious Learner | Restless Dreamer

Don't learn AI - ask questions instead. Take the doc (link in first comment) with a list of 20 important questions, add your company/app name to get a rich understanding of how AI can help you in your context. -- I’ve spent the last few weeks immersed in a deep dive into GenAI System Design; following are the key learnings: (1) AI Foundations >> AI Tools (2) Going deeper into 1 problem >> Following AI news & getting overwhelmed (3) Mind-map 👇 >> reading / memorizing everything around AI (4) Debugging is a feature, not a bug (5) Post-launch iterations is 80% work (now that launching is simple) -- Answering frequently asked questions: (1) If Claude Code can execute everything, why do we need to learn AI? You don't, if you are OK with an average output. The output of an un-optimized AI pipeline will feel similar to wearing your shoes on the wrong feet 👞: it technically gets you to the mailbox, but every step is a silent scream for help. (2) Will all of us loose our jobs? If you are still asking this question in Mar 2026, you are a by-stander watching the show. Instead, join the show and you will automatically be able to answer your own question. Pick a AI-solvable problem and solve it - could be a simple email digest for your-self, or a new startup idea! It will make you "AI-native" which is the best way to escape the "fear of job loss". (3) Can a non-coder learn / build? I had a former pilot ✈️ in my course's cohort. Also had a VP at McKinsey. Also had a 58 year old supply chain professional. I personally also have never written a single line of code. Yes - its possible and not that difficult either. If it feels intimidating, thats good because that will become your edge (everyone else is also feeling intimidated) (4) Starting to learn / build using AI feels overwhelming. Where to begin? ⭐ Quest for "perfection" adds un-intended stress. Start somewhere. ⭐ Using the most expensive tools/stack to start building might cost less than $30 per month for a side-project. Focus on effectiveness and save efficiency for later. (when you start, you are mostly within free tiers) ⭐ If you were learning basketball, watching Jordan might be overwhelming. But this sport has no experts yet. So - no fear of judgement! "Learn/build AI products" = "Jordan of AI" (5) So much content out there - whats the best way to learn? ⭐ Asking this question means you are trying to be efficient --> "I want most value in minimum price" ⭐ Pick a problem and start building; learning will happen as a by-product. -- Would highly recommend the following course: (1) Lifelong access to slack community (2) Lifelong access to all study material which will keep getting updated with newer cohorts (3) Incredible support to answer questions / engage in discussions (sync + async) Course: Building Agentic AI Applications with a Problem-First Approach (lnkd.in/g7DHahnD) Thanks for the amazing course to : Kiriti Badam Aishwarya Naresh Reganti LevelUp Labs

Don't learn AI - ask questions instead. Take the doc (link in first comment) with a list of 20 important questions, add your company/app name to get a rich understanding of how AI can help you in your context. -- I’ve spent the last few weeks immersed in a deep dive into GenAI System Design; following are the key learnings: (1) AI Foundations >> AI Tools (2) Going deeper into 1 problem >> Following AI news & getting overwhelmed (3) Mind-map 👇 >> reading / memorizing everything around AI (4) Debugging is a feature, not a bug (5) Post-launch iterations is 80% work (now that launching is simple) -- Answering frequently asked questions: (1) If Claude Code can execute everything, why do we need to learn AI? You don't, if you are OK with an average output. The output of an un-optimized AI pipeline will feel similar to wearing your shoes on the wrong feet 👞: it technically gets you to the mailbox, but every step is a silent scream for help. (2) Will all of us loose our jobs? If you are still asking this question in Mar 2026, you are a by-stander watching the show. Instead, join the show and you will automatically be able to answer your own question. Pick a AI-solvable problem and solve it - could be a simple email digest for your-self, or a new startup idea! It will make you "AI-native" which is the best way to escape the "fear of job loss". (3) Can a non-coder learn / build? I had a former pilot ✈️ in my course's cohort. Also had a VP at McKinsey. Also had a 58 year old supply chain professional. I personally also have never written a single line of code. Yes - its possible and not that difficult either. If it feels intimidating, thats good because that will become your edge (everyone else is also feeling intimidated) (4) Starting to learn / build using AI feels overwhelming. Where to begin? ⭐ Quest for "perfection" adds un-intended stress. Start somewhere. ⭐ Using the most expensive tools/stack to start building might cost less than $30 per month for a side-project. Focus on effectiveness and save efficiency for later. (when you start, you are mostly within free tiers) ⭐ If you were learning basketball, watching Jordan might be overwhelming. But this sport has no experts yet. So - no fear of judgement! "Learn/build AI products" = "Jordan of AI" (5) So much content out there - whats the best way to learn? ⭐ Asking this question means you are trying to be efficient --> "I want most value in minimum price" ⭐ Pick a problem and start building; learning will happen as a by-product. -- Would highly recommend the following course: (1) Lifelong access to slack community (2) Lifelong access to all study material which will keep getting updated with newer cohorts (3) Incredible support to answer questions / engage in discussions (sync + async) Course: Building Agentic AI Applications with a Problem-First Approach (lnkd.in/g7DHahnD) Thanks for the amazing course to : Kiriti Badam Aishwarya Naresh Reganti LevelUp Labs

Mar 12, 2026

Kunal Tiwary

Data Engineering | Machine Learning | Artificial Intelligence | Engineering Leadership

For a long time, I’ve believed in fundamentals — regardless of the problem at hand. That belief has served me well. It’s helped build the muscle to tackle new domains with clarity and design solutions that are grounded rather than reactive. That principle hasn’t changed in the age of AI. If anything, it matters more — especially with so much noise around tooling and far less discussion about the underlying problem. That’s one of the reasons I decided to take Building Agentic AI Applications with a Problem-First Approach by Aishwarya Naresh Reganti and Kiriti Badam via Levelup Labs What I value about this course isn’t that it presents something radically new. It reinforces disciplined thinking in a space that often rewards premature complexity. When building AI systems, it’s easy to jump straight to: • Should this be an agent? • Do we need RAG? • How do we orchestrate multiple tools? But the sharper questions are: • What exact decision are we improving? • Which constraints truly matter? • What does failure look like in this context? AI doesn’t replace fundamentals. It amplifies weak ones. This course has been a strong reminder that starting with the problem — clearly and rigorously — isn’t optional just because we’re working with LLMs. If anything, the cost of skipping that step is higher. Sharing the link for anyone interested in approaching AI systems with systems thinking rather than simply stacking components: lnkd.in/gJgg82-d Creators: Aish – Aishwarya Naresh Reganti (x.com/aishwarya_k) Kiriti – Kiriti Badam (x.com/kiritikh) I’ll share one specific shift in how I’ve been approaching agent design as a result of this reframing in my next post. #AgenticAI #SystemDesign #GenAI #LevelUpLabs #EngineeringLeadership

Mar 12, 2026

Nikhil Jakkani

Performance Engineering

Over the past 6 weeks, I completed “Building Agentic AI Applications with a Problem-First Approach” with Aishwarya Naresh Reganti and Kiriti Badam—a game-changer for navigating today’s GenAI landscape. In a world of nonstop AI hype, this course helped me focus on what matters for the enterprise: real-world system design, rigorous evaluation, and agent orchestration. The vibrant cohort stood out—collaborating with peers accelerated my learning and made the experience that much richer. Big thanks to Aishwarya and Kiriti for their clarity and technical depth. My capstone project, AWSentinel, let me apply these concepts: thinking from a systems lens, iterating and testing, and building an agent-powered AWS compliance and governance platform with valuable feedback throughout. If you’re looking to build practical GenAI skills and join a strong community, I highly recommend this course.

Dec 11, 2025

Meera Vinod

AI Product Manager | Responsible AI and HCI | UC Berkeley Alum

Really enjoyed taking the course on Building Agentic AI applications by Aishwarya Naresh Reganti and Kiriti Badam. Even though I had hesitations about taking a flipped class model course in the beginning, I soon started enjoying the in-person sessions where we dove into the material and the technical discussions. I also really appreciated the thought they put into helping structure our final projects, which was the most enjoyable part for me. I really liked the emphasis on solving actual problems with AI even though it was meant to be a capstone project to finish the course. Overall, it was a very rewarding experience racing to the finish line with my team mates Mahtab Soin, Manpreet Arora, Ameya Bongale, Faith Morante, Antoaneta V. working on some very challenging problems in education. Appreciate the learnings, and the cameraderie formed! For anyone interested in the course, link in the comments.

Nov 17, 2025

NM

Niranjan Mulay

Sr. ML Engineer | AI/ML Product Development | Ex-PM Technical (Edge AI & Multimedia)

The Agentic AI space is exploding, and it's hard to keep up with all the buzz. That's exactly why focusing on fundamentals matters. I've been quietly following the work of Aishwarya Naresh Reganti and Kiriti Badam for a while now—their insights have consistently stood out. So when they launched their course "Building Agentic AI Applications with a Problem-First Approach", I decided to take the plunge. What makes it exceptional is its 3-phase structure (Core, Build, Grow) that takes you from concepts and frameworks to actually building agentic AI applications—all with their trademark "problem-first & iterative building" approach. It strikes a rare balance between breadth and depth, clearly born from their hands-on experience and genuine passion for educating the community. More importantly, it doesn't just present facts and resources—it encourages you to think and operate at a layer beneath those facts, whether you're building workflow agents, agentic RAG applications or ReAct patterns. If you're looking to build a solid foundation in this space, check out the next cohort: lnkd.in/g4PxmUWB

Nov 17, 2025

V

Vishnu S.

Senior Data Scientist at Humana

I recently completed the "Building Agentic AI Applications with a Problem-First Approach" course by Aishwarya Naresh Reganti and Kiriti Badam and I wanted to share my experience. The curriculum is built around a problem-first approach. Rather than focusing on the rapidly changing technology stack and tools, it encourages you to step back and evaluate the core business problem. The development philosophy is highly pragmatic. The course emphasizes iterative design, starting with a simple solution and refining it over time, rather than attempting to build a complex system from day one. The course covered a wide range of topics from RAG systems and prompt engineering, MCP to agent frameworks, evaluation strategies, and production deployment considerations. It also gave us the opportunity to build a use case as we go and really think about system design considerations more than the tool utility. While the core sessions lay a solid foundation, the course still provides more immense content and additional reading material that I will keep revisiting. One of the best parts - the community. Right from startup founders, software engineers, product managers, solution architects, and data scientists, you name it — the space was constantly buzzing with ideas, questions, and support. Witnessing hundreds of participants move from initial ideation to delivering working AI systems within just six weeks was a testament to the course's structure and support. Highly recommend the course - you can find the details here: lnkd.in/gmWVbG4u

Nov 17, 2025

Anitha Abraham Kuriakose

Recently wrapped up six incredible weeks with the “Building Agentic AI Applications with a Problem-First Approach” course by Aishwarya Naresh Reganti and Kiriti Badam, and it’s been truly transformative for me. When I started, terms like RAG, prompt engineering, evals, guardrails, and optimizations felt like a foreign language but now I not only understand them, I can confidently use them to design and discuss AI workflows. The course was hands-on from day one, shifting my mindset from chasing the latest models to building AI solutions for real problems. I learned how powerful prompt engineering can turn a vague idea into a clear, practical AI output, and discovered how RAG enriches AI with up-to-date, domain-specific knowledge. The focus on evaluating systems, implementing guardrails, and optimizing for cost and latency was game-changing. The final sprint which was building an AI mortgage consultant in just one week pushed me to synthesize all this learning in record time. Thank you to my amazing teammates - Srejith Ramesh Isaac Lionel Luis Andres Ortega Orduño Soumya Kanti Banerjee Sarath Satheesan - our collaboration, brainstorming sessions, and those late-night “solution-finding” marathons taught me as much as the formal course did. To anyone considering this course: lnkd.in/eYdzw4vN If you want a practical, community-driven experience that turns technical jargon into real skills and if you aren’t afraid to work hard with a diverse, supportive cohort, this is for you.

Nov 17, 2025

Vivek Sundaram

Product & Strategy l Rocketlane, Dream Sports, CaratLane, Hopscotch | IIMB, NITT

I recently completed my first AI course, and it's been transformative. While exploring ways to learn AI fundamentals and its applications practically, I came across many courses but none quite clicked. Then I saw Gagan Biyani's shoutout on LinkedIn about this course. I checked it out, loved the structure, and enrolled immediately. A huge thank you to Aishwarya Naresh Reganti and Kiriti Badam for their exceptional teaching and skill in building strong core fundamentals. The course was practical, hands-on, and incredibly impactful. Course Link:

Nov 17, 2025

Balaji Raghunathan

Data & AI Engineering Business Unit Leader at Sigmoid

Happy to have obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven! This course lnkd.in/gvj2Eyvg is for those looking at exploring the intersection of Business Value, System Design and Agentic AI. Thanks to the way Aishwarya Naresh Reganti and Kiriti Badam articulated the concepts and designed the course while blending in practicality and their experience lnkd.in/gR7GDAKM

Nov 17, 2025

Venkat V.

Global Engineering Leader at Riverbed Technology

𝗣𝗮𝘁𝗶𝗲𝗻𝘁𝘀 𝗳𝗲𝗮𝗿 𝗵𝗼𝘀𝗽𝗶𝘁𝗮𝗹 𝗯𝗶𝗹𝗹𝘀 𝗺𝗼𝗿𝗲 𝘁𝗵𝗮𝗻 𝗽𝗿𝗼𝗰𝗲𝗱𝘂𝗿𝗲𝘀. That became my north star for the last few weeks. I just completed the Maven “𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜 𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝘄𝗶𝘁𝗵 𝗮 𝗣𝗿𝗼𝗯𝗹𝗲𝗺-𝗙𝗶𝗿𝘀𝘁 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵”. It was exceptional. Huge credits to Aishwarya Naresh Reganti and Kiriti Badam for an outstanding course design. Their program pushes you to think deeply about non-deterministic systems, responsible design principles, evaluation frameworks, and the realities of building AI that functions in complex domains. The lectures and guest sessions on cost/latency trade-offs, optimization strategies, guardrail design, evaluation metrics, and scaling prototypes into production-ready systems were especially noteworthy. They didn’t just teach tools or techniques. They taught a mindset. Along with Julian K. & Pankhuri Singhal, I built an early version of Care Compass, an AI system that reduces a traditionally long, multi-step process to bring clarity regarding patient's coverage and costs 𝘣𝘦𝘧𝘰𝘳𝘦 they seek care. The system evolved across 3 iterations: → Iteration 1: Text-based reasoning → Iteration 2: Human-in-the-loop guardrails → Iteration 3: Multi-agent workflows The result: A working prototype that compresses 5+ weeks → 25 minutes. Over the coming months, I’ll be continuing this work exploring the ecosystem, stress-testing assumptions, and validating what’s possible: • How financial clarity shapes patient decisions • Where administrative friction adds delay • How hospital and insurance processes interact • Where operational bottlenecks create avoidable anxiety • How AI can support better outcomes across stakeholders If you’re working in healthcare, insurance, hospital operations, patient experience, or AI agents, I’d love to connect. Your questions, feedback, or collaboration could shape where this exploration leads. Read more about the course here: lnkd.in/gWizek7S #HealthcareInnovation #AIinHealthcare #PatientExperience #MavenLearning #HealthTech

Nov 17, 2025

JP

Joshua Pelletier

Senior Product Manager at Truepill

I spent the last few weeks in Building Agentic AI Applications with a Problem First Approach with Aishwarya Naresh Reganti and Kiriti Badam I went in to learn about LLMs. I came out with new teammates. Not real teammates. New ways of working that changed how I build. The Researcher helped me slow down and sharpen the problem. The Evaluator pushed me to show proof and build the plan first. The Builder wanted action and wanted one more idea every day. I had to guide that one away from feature creep. I am a Product Manager in the Loop. Together we built a working system. A full CRAG flow with retrieval and refinement. A judge model for semantic scoring. A ground truth set for validation. An instrumentation layer that showed how each step behaved. Real tools. Real code. Real learning. Aish shared an idea that stayed with me. Language is the interface for these systems. No buttons. No menus. Just words. The way we speak to them shapes the work they do. After this course AI feels less like magic and more like a team. A brilliant and chaotic team that needs clear problems and clear measures. I feel energized by this shift and grateful for the work. If you are thinking about investing in your future do not wait. lnkd.in/gf5VunKa Excited to bring these new teammates into my efforts to build trustworthy AI in enterprise. #AgenticAI #AIInHealthcare #ProductManagement #Evaluation #DigitalHealth

Nov 17, 2025

DURGA PRASAD

Conversational AI and MEAN Stack Developer at Eli Lilly and Company

Just wrapped up an amazing learning experience with the course “Building Agentic AI Applications with a Problem-First Approach.” What made this experience stand out? ✅ A refreshing focus on real-world enterprise AI—not chasing trends or hype, but solving meaningful business problems with clarity and purpose. ✅ Practical strategies for designing enterprise agentic AI systems that scale and deliver measurable impact. ✅ Hands-on learning: Iterative building of solutions through assignments and discussions, not just passive video watching or boring lectures. ✅ An incredible community of professionals sharing insights, challenges, and ideas—learning together was the best part! "𝗔𝗜 𝗶𝘀 𝗮 𝘁𝗼𝗼𝗹, 𝗻𝗼𝘁 𝗺𝗮𝗴𝗶𝗰. 𝗪𝗵𝗲𝗻 𝗮𝗽𝗽𝗹𝗶𝗲𝗱 𝘁𝗵𝗼𝘂𝗴𝗵𝘁𝗳𝘂𝗹𝗹𝘆, 𝗶𝘁 𝘁𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝘀 𝗼𝗿𝗴𝗮𝗻𝗶𝘇𝗮𝘁𝗶𝗼𝗻𝘀." If you’re serious about moving beyond buzzwords and building enterprise AI solutions that matter, 𝗜 𝗵𝗶𝗴𝗵𝗹𝘆 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱 𝘁𝗵𝗶𝘀 𝗰𝗼𝘂𝗿𝘀𝗲. Big thanks to Aishwarya Naresh Reganti and Kiriti Badam for creating such a high-impact experience! #EnterpriseAI #AgenticAI #ProblemFirstApproach #ContinuousLearning #LeadershipInTech #Maven #AILeadership #AppliedAI

Nov 17, 2025

chetan karande

@karande_c

An excellent course on designing and building impactful Agentic AI Systems by Aishwarya Naresh Reganti and Sai Kiriti Badam maven.com/aishwarya-kiriti/genai-system-design. Along with techniques, it offers a practical approach to think about Agentic AI in a problem-first manner.

Nov 15, 2025

Uttarika Shetty

Founder | Product Manager | AI & Learning

I completed the Agentic AI course: lnkd.in/g7qGKNTx and would like to thank Aishwarya Naresh Reganti and Kiriti Badam for creating such an incredible program! Some things that stood out to me: 1. The program's content is incredibly comprehensive, offering a depth and quality that allows for continuous learning with each revisit. It evolves alongside your understanding, ensuring a rich and engaging learning experience. 2. The inclusion of industry experts as guest speakers was a highlight, providing valuable insights into the nuances of various AI roles and their real-world applications. 3. A special mention goes to the team's responsiveness, fostering a vibrant learning community. The active engagement through detailed responses, peer discussions, and knowledge sharing on Slack helped create a space where I learned more than I could imagine! I am thrilled to implement the knowledge gained from this course and continue to grow in the field of Agentic AI. 🚀

Nov 14, 2025

Abhishek Sinha

Product Leadership | MBA (SPJIMR)

It’s been an incredible experience completing an intensive, structured six-week journey through the "Building Agentic AI Applications with a Problem-First Approach course". Right from building a workflow agent in LangGraph to building Agentic RAG & multi agent systems, using tools via MCP and adding memory and persistence to my AI system, the entire journey has been deeply rewarding . Some of my major learnings from the course about AI systems have been 🎲 Managing Non Determinism Every AI system is non deterministic in nature. Be frugal in adding any additional component as the non determinism increases with every component addition. ⚙️ Optimizations Start with a very basic AI system, optimize for prompts, optimize for RAG and only if your Evals are not being met then move towards more complicated systems like Agentic RAG or Multiagent systems. 📊 System Design & Evals Always do a first cut of systems design yourself & be actively involved with the engineering team over multiple iterations of system design. Evals are required for each and every component of an AI system to meet your end goal or metric. 🔎 Observability Monitoring, tracing every stage of an AI system via Comet Opik. 🛡️ Guardrails Implementation of Guardrails for all the safeguards, constraints and policies an enterprise might have. Thank you Aishwarya Naresh Reganti & Kiriti Badam! It is a very well thought out and carefully designed course, it takes you from 0->10 and from 10->100 pretty quickly provided you put in sufficient effort side by side. The support for course content & hands on exercises has been incredible. Some of the guest lectures gave me insights on actual enterprise grade implementations and the challenges they have faced while implementing an AI system. Thank you Sahana Venkatesh, Ashwin Naidu for all your support. lnkd.in/gz-bMevy

Nov 14, 2025

Chandrasekhar Muthyala

AI Engineer | Agentic AI | Senior Software Engineer | ReactJS | GCP certified

✨ Grateful for an Incredible Learning Journey! ✨ I completed my first AI course, and it has been an amazing experience! A big thank you to Aishwarya Naresh Reganti and Kiriti Badam for their extraordinary skills in building strong core fundamentals that made this journey so impactful. 🌟Master the fundamentals, unlock the future!🌟 It was a pleasure partnering and collaborating with my excellent team on the Capstone Project, which gave me deep insights and knowledge through multiple iterations. Special thanks to my amazing teammates: Sujana Pinjala Pritish Dhavan, Usha Kasiraman, Shanti Chilukuri, Sreeni Kanna, Juan Carlos Benayas Arroyo. Course Link: lnkd.in/gWizek7S

Nov 14, 2025

Birinder Singh

AI & Data Science Lead | UBC | IIT Roorkee

From prompt engineering to agentic design and a Runner-up capstone finish. Pleased to share that I’ve completed the cohort-based program “Building Agentic AI Applications with a Problem-First Approach”, by Aishwarya Naresh Reganti and Kiriti Badam, and championed by Sahana Venkatesh and Ashwin Naidu. The learnings were immense. Having followed Aish and Kiriti for over a year, it was great to finally learn from them firsthand. The lens of Evaluations, Guardrails, Cost, Latency, and Effort gave me a deeper appreciation for how enterprise-grade AI systems need to be designed. With an evaluation-driven and iterative design mindset, I now see even greater opportunities to amplify the impact of my Generative AI work. Putting all these learnings into practice, our capstone Reva – Revenue AI (an Agentic Pricing Strategy Assistant) was recognized as Runner-up, narrowly missing the top spot, where we tackled a real pain point in how enterprises manage pricing strategies in dynamic and price-sensitive markets. Through three iterative design cycles, we evolved from user-driven to autonomous, with a single agent generating recommendations validated by human feedback. At the intersection of Generative AI, Machine Learning, and statistical elasticity modelling, we identified clear signals of product potential and proposed A/B tests as a structured approach for empirical validation. After developing AI-powered discovery and recommendation frameworks and now prototyping an Agentic Pricing Strategy Assistant, I’ve further deepened my understanding of bridging AI theory to enterprise-scale applications. Excited to translate these insights into real business impact at Rocket Rocket Innovation Studio #AgenticAI #GenerativeAI #MachineLearning #ContinuousLearning

Nov 14, 2025

Vikram Moorjani, MBA

Product & Technology Leadership | Advisor | E-commerce [B2B & B2C] | Developer Productivity Apps

Proud that our Team Athena won the Capstone project competition for the October Cohort for Building Agentic AI Applications with a Problem-First Approach [lnkd.in/gr5zNEJt]! As I continue to deepen my AI product expertise, I have focused on discovery, technical guardrails, and evaluation frameworks for our government contract-matching solution. Collaboration with Puja Potdar on roadmap and scope challenges reinforced the critical role of critical thinking in separating successful AI initiatives from feature bloat. Abhi🔹 M. - stupendous work on n8n workflows and architecture. Amit Mukherjee - enjoyed collaborating on the UI and appreciated your deep technical expertise. Liz Filardi did an excellent job on the Loom video and deck, keeping us on track. Randy Sauers, PMP - thanks for the original concept. What we built in a week typically takes teams months. It is a testament to how each team member put their best foot forward. The team came together, and that's what matters. Most significant learning: Keep it simple and, as the course suggests, "Problem First". AI makes feature creep easy and affects cost/performance. We cut inference by 83% with LPUs and learned to manage context windows for large payloads. All the teams that participated gave up their time; you are all winners. October Cohort, take a bow. To our course leaders, Aishwarya Naresh Reganti and Kiriti Badam, thank you for all the hours you have spent designing the course. So much learning in such a short time, I would love to come back to learn again. #AI #ProductManagement #AgenticAI #GovTech

Nov 14, 2025

Luis Andres Ortega Orduño

Site Reliability Engineering - Director en FICO®

I just obtained certificated from Maven Building Agentic AI Applications with a Problem-First Approach. Thanks to Aishwarya Naresh Reganti and Kiriti Badam for putting together this course which is impressive on how much support you get from the community they have created. This is actually a movement itself.

Nov 14, 2025

Usha Kasiraman

Complex Systems Analysis | Solutions Design | Agile | Continuous Learner

My AI learning journey started a month ago when I joined the Problem First AI Course run by Aishwarya Naresh Reganti and Kiriti Badam on Maven. The course takes you from foundational AI concepts to building advanced Agentic AI flows. I have learnt so much in the past month which I could not learn my own in the past year. Aishwarya's core concept videos are concise and packed with knowledge at the right level. She demystifies the AI jargons and systematically builds the conceptual knowledge. (LLMs, Prompting, Evals, Tokens, Agents etc.) Learning to build on LangFlow (being non-technical) is truly empowering. Kiriti does a great job of explaining the build assignments, clarifying the doubts and keeping it challenging throughout. My big takeaways from the course are: -the clarity on how LLMs are used -understanding how prompting plays a big part in the LLM's response, -RAG concepts -importance of Evals -building the flows itself My "aha" moment was to see how non-deterministic systems respond and the mind shift needed to develop Gen AI applications. The ideas and concepts from the course will resonate with all AI practitioners - I really enjoy the opportunity to interact with other professionals. Thank you so much Aishwarya Naresh Reganti and Kiriti Badam ! Check out the course Building Agentic AI with a Problem First approach on Maven platform!

Nov 14, 2025

Vibha R.

AI Enthusiast and Tech Optimist | Engineering Leader at MSCI | Ex-Microsoft, Ex-Meta

I recently attended this course on Maven - Problem First AI (lnkd.in/etidfdNk) taught by two amazing instructors Aish(lnkd.in/e_9M2KYj) and Kiriti(lnkd.in/e5WEbqaG). They cover how to go deep into choosing and fine-tuning models on multi-modal data, agentic vs workflow use-cases, and running evals - overall, how to build AI systems that solve real enterprise use cases. It has high-level theory, hands-on coding, talks on the latest in AI and a wonderful community to continue learning this space together. Highly recommend! Its one of the most popular AI courses on Maven for a reason and also featured in Lenny's list.

Nov 14, 2025

Reji Ravindran

Program Manager @ Ivanti

Completed this program recently: lnkd.in/gz5UyirV This is not a traditional course—it's an experience in deeply distilled knowledge. For senior professionals evolving beyond tactical proficiency, this program delivers strategic clarity. Years of enterprise experience are distilled into key ideas with exceptional signal-to-noise ratio. Every hour yields maximum return. The curriculum focuses on core concepts and real-world experience, not fleeting tools. In an industry of transient frameworks, it provides foundational principles. It teaches the why and how, equipping you to evaluate future technologies with clarity. The empathy and collaboration fostered by instructors and community elevate the experience. This program functions as a foundational text that deepens with every read. Each revisit reveals new layers and strengthens your grasp of complex relationships. Highly recommended for leaders building durable, scalable solutions. Aishwarya Naresh Reganti, Kiriti Badam - You have started a movement here!

Nov 14, 2025

Dewan Hussain

Staff Software Engineer focused on C#, Dotnet Core, and Microservices

Just completed the Maven Course "Building Agentic AI Applications with a Problem-First Approach." A big thank you to Aish and Kiriti Badam for organizing this course. While some content was familiar to me from before the course, I gained a lot of formal learning throughout the sessions. What I appreciated most were the industry experts who shared their insights on solving real-world problems. Additionally, the Chai sessions were always filled with pleasant surprises. I truly value the effort and passion that went into designing this course. Thank you Ravi Kiran Nukala for suggesting this course. It was totally worth it. #blackline #generativeai #maven

Nov 14, 2025

Murali Reddymalla

Principal Project Manager | Leading Oracle CC&B, SAP S4 HANA & Cloud transformations | Execution Excellence

I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven! You know that you need to start somewhere to be part of the AI evolution in a meaningful way, but given so much noise and hype, where do you get started? I am fortunate to find the course run by experienced and passionate instructors Aishwarya Naresh Reganti and Kiriti Badam who are not just educating but building a community that you can carry beyond the course. Here are my top 10/10 takeaways that made this course truly special: 1. Exceptional Instructors — Aishwarya Naresh Reganti and Kiriti Badam are not just knowledgeable, but deeply passionate AI leaders. Their energy, clarity, and curiosity made every session engaging and thought-provoking. 2. The Problem-First Mindset — A brilliant concept that shifts focus from chasing tools to identifying meaningful problems worth solving using AI. 3. Beautifully Structured Content — The course moves from fundamentals to advanced concepts seamlessly, helping you connect the dots between theory, practice, and innovation. 4. Agentic AI — Beyond the Buzzword — I finally understood what it truly means to design intelligent, adaptive systems that think in context rather than follow static patterns. 5. Real-World Relevance — The examples, discussions, and frameworks are directly applicable to enterprise and product scenarios, making the learning instantly useful. 6. Guest Lectures that Inspire — Listening to professionals building AI systems brought immense clarity to what’s happening in the industry right now. 7. Collaborative Learning Community — The slack discussions and peer projects created a safe space to explore ideas, ask questions, and learn from each other. 8. Guided Experimentation — It wasn’t just theory — every concept encouraged us to build, test, and reflect. That hands-on aspect made all the difference. 9. No Hype, Only Depth — The course stays away from buzzwords and focuses on practical, grounded knowledge — something I deeply value. 10. Lasting Impact — This wasn’t just another course; it changed how I think about problems, systems, and innovation itself. Truly a 10/10 experience. Thank you, Aishwarya Naresh Reganti and Kiriti Badam, for curating such a meaningful learning journey. Highly recommend this to anyone who wants to build with intention and purpose in the world of AI. #AI #ProblemFirst #GenAI #AgenticAi #EnterpriseAI #LLM #ContinuousLearning

I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven! You know that you need to start somewhere to be part of the AI evolution in a meaningful way, but given so much noise and hype, where do you get started? I am fortunate to find the course run by experienced and passionate instructors Aishwarya Naresh Reganti and Kiriti Badam who are not just educating but building a community that you can carry beyond the course. Here are my top 10/10 takeaways that made this course truly special: 1. Exceptional Instructors — Aishwarya Naresh Reganti and Kiriti Badam are not just knowledgeable, but deeply passionate AI leaders. Their energy, clarity, and curiosity made every session engaging and thought-provoking. 2. The Problem-First Mindset — A brilliant concept that shifts focus from chasing tools to identifying meaningful problems worth solving using AI. 3. Beautifully Structured Content — The course moves from fundamentals to advanced concepts seamlessly, helping you connect the dots between theory, practice, and innovation. 4. Agentic AI — Beyond the Buzzword — I finally understood what it truly means to design intelligent, adaptive systems that think in context rather than follow static patterns. 5. Real-World Relevance — The examples, discussions, and frameworks are directly applicable to enterprise and product scenarios, making the learning instantly useful. 6. Guest Lectures that Inspire — Listening to professionals building AI systems brought immense clarity to what’s happening in the industry right now. 7. Collaborative Learning Community — The slack discussions and peer projects created a safe space to explore ideas, ask questions, and learn from each other. 8. Guided Experimentation — It wasn’t just theory — every concept encouraged us to build, test, and reflect. That hands-on aspect made all the difference. 9. No Hype, Only Depth — The course stays away from buzzwords and focuses on practical, grounded knowledge — something I deeply value. 10. Lasting Impact — This wasn’t just another course; it changed how I think about problems, systems, and innovation itself. Truly a 10/10 experience. Thank you, Aishwarya Naresh Reganti and Kiriti Badam, for curating such a meaningful learning journey. Highly recommend this to anyone who wants to build with intention and purpose in the world of AI. #AI #ProblemFirst #GenAI #AgenticAi #EnterpriseAI #LLM #ContinuousLearning

Nov 14, 2025

AM

Amit Mukherjee

Principal Cloud Solutions Architect

Just wrapped up an incredible week building an AI-powered Contracting Opportunities Assistant as our capstone project for the Problem First AI course! Working with Liz Filardi, Abhi🔹 M., Puja Potdar, Vikram Moorjani, MBA, and Randy Sauers, PMP across different timezones was exhilarating. We pivoted, iterated, and came out stronger by genuinely supporting each other and incorporating feedback throughout. Honored that our project resonated with the cohort, but what really mattered was the "Problem First + Ego at the Door" mindset that Aishwarya Naresh Reganti and Kiriti Badam along with Sahana Venkatesh and Ashwin Naidu constantly championed. The coolest part was we learned and deployed tools like v0, n8n (course supported Langflow, so kudos to you Abhi🔹 M.), and Supabase thoughtfully, not just using tech for tech's sake, but solving real problems through iteration. We even mapped out a full product roadmap and captured key lessons learned along the way, from optimizing LLM inference times to managing API costs smartly. The course's framework (Core: Learn First, Build: Create Next, and Grow: Cultivate Mindset) intelligently delivered on making everyone AI-wise in just 6 weeks, regardless of background setting up an excellent foundation. If you're looking to get genuinely AI-grounded and build real applications, check out the course: lnkd.in/eAmhYUZb Grateful for this experience and the supportive community around it.

Nov 14, 2025

Mahtab Soin

Building Master Ed

One of the highest leverage activities I engaged in this year was taking Aishwarya and Kiriti's excellent course on Maven called "Building Agentic AI Applications with a Problem-First Approach". As someone building AI-first products in education, the course offered exactly what I needed: a top-down approach with a holistic systems-thinking foundation, followed by deep technical and empirical nuance. There are so many levers to architect and optimize in agentic AI, and for such a fast-moving field, this course has been both a worthy investment and a massive unlock. I've applied many of these learnings directly to Master Ed - more exciting updates coming soon! Huge thanks to Sahana and Ashwin for their outstanding support, and to my capstone team Faith, Meera, Antoaneta, Ameya and Manpreet for their contributions to Master Ed.

Nov 13, 2025

Anurag Sharma

Director of Engineering | Enterprise Architect |Data Science and Machine Learning| Generative and Agentic AI

Excellent learning experience. The course equips you with solid fundamentals, practical techniques, and the confidence to build—plus, clear strategies for taming LLM ‘chaos. lnkd.in/gw--Z-N7 Aishwarya Naresh Reganti Kiriti Badam

Nov 13, 2025

IL

Isaac Lionel

Designing AI Experiences at Lloyds Banking Group

“To truly own knowledge, you must argue with it, question it, and leave your intellectual fingerprints all over it.” - Mortimer Adler What an incredible 6 weeks it has been! Intense, jam-packed, collaborative, challenging, and yet deeply rewarding. I’m thrilled to share that I’ve completed the "Building Agentic AI Applications with a Problem-First Approach” certification. (lnkd.in/eAeWfQnf) I enrolled in this course as part of my 2025 self-learning goals, to better understand what really happens behind the scenes of Agentic AI systems and why they behave the way they do. I’ll admit, at first I was anxious (the classic imposter syndrome creeping in), wondering if I had made the right choice. But the way this course was structured (with the perfect blend of theory, collaboration, and hands-on learning) quickly turned that anxiety into excitement and confidence. This wasn’t just another “watch-and-learn” course. It was a space for thinking, building, and questioning. While there’s plenty of “Agentic AI” content scattered across YouTube and blogs, most stay at the surface. This course drilled deep into how to think about building Agentic AI applications - problem-first. Understand the system so deeply that you can have fun with it, without breaking it. This course could not have been designed without immense passion and deep knowledge of this space. Aishwarya Naresh Reganti and Kiriti Badam have truly nailed it, creating a rich and transformative learning experience. All I can say is: Thank you 🙏 for crafting this journey. I’m incredibly grateful for the experience and the amazing community that came with it. Here’s to continuously learning, unlearning, and reimagining what’s possible with AI. #AI #AgenticAI #GenAI #ContinuousLearning #SystemDesign #AIApplications #LifelongLearning

Nov 13, 2025

GC

Gilda Fernandez Concha

Data Metrics Engineer @ UBS

Excited to share my certificate of “Building Agentic AI Applications with a Problem-First Approach” from Maven. I highly recommend this course to anyone interested in learning how to design AI agents that solve real business problems. Thank you very much to Aishwarya Naresh Reganti and Kiriti Badam for their clear guidance and for creating an inspiring, collaborative environment that encouraged critical thinking and experimentation. Proud that our capstone project, “Hack for a Saturated Real Estate Market” came very close to first place. Together with Cosima Lefranc, Thaung Su Nyein, Sheriff Shitu and Ankit Mathur, we worked across three continents to design an AI agent tackling Switzerland’s competitive rental market. #AI #AgenticAI #ArtificialIntelligence #Maven #ProblemFirst #Innovation

Nov 13, 2025

Anish P.

Software Engineering Senior Manager

lnkd.in/eVR5AU7f Thrilled to have completed the "Building AI Agentic Applications with a Problem-First Approach" program! A huge thank you to Aishwarya Naresh Reganti and @https://lnkd.in/eTkqZGmj for crafting such a well-rounded and inspiring journey. This experience stood out for its vibrant community, thoughtfully structured learning path, and the opportunity to dive into a capstone project that brought theory into practice. The post-course support was a meaningful bonus—something that truly set this program apart. The three-phase learning path—Core, Grow, Build—was brilliantly designed: Core laid the foundation with essential meta-concepts, helping me grasp the broader agentic landscape. Grow offered a curated set of high-quality resources, cutting through the noise of online content and guiding deeper exploration. Build was all about application, with both low-code and full-code tracks that made the experience inclusive and hands-on.

Nov 13, 2025

Priyank Agrawal

Software Engineering Manager, VP @Chase | Data Analytics | Incoming MSCS @ GaTech

Before joining the course, I considered myself AI-illiterate. After completing it, I can confidently say that I’m now AI-literate. There is still a lot more to learn, but this course gave me a solid foundation and introduced meta-concepts that I can continue to build on. The learning path—Core, Grow, and Build—was especially well designed. • Core focused on the essential meta-concepts, giving me the structure I needed to understand the broader landscape. • Grow curated high-quality resources to deepen understanding. This was particularly valuable because there is an overwhelming amount of material online, and having a guided, intentional set of resources is something I’ve rarely seen in other courses. • Build provided the space to apply what we learned. I appreciated that the course offered both low-code and full-code tracks, making it accessible to learners with different technical backgrounds. I also want to acknowledge the coaching and staff. They were consistently available and responsive—not only to questions directly related to the content and assignments, but also to broader, field-level questions. Their willingness to share their own experiences added meaningful context to the learning process. Overall, the course was structured, thoughtful, and practical, and it significantly accelerated my understanding of AI. Big thanks to Aishwarya Naresh Reganti Kiriti Badam and the entire team behind this course.

Nov 13, 2025

Sri Neeharika Kollipara

Data Scientist @ Lenovo | NEU | Ex Data Analyst @ Arrow Electronics | Ex Cloud Analyst @ Oracle

🚀 From learning prompts to building agents — and winning 1st place doing it! 🏆 I recently completed the “Building Agentic AI Applications with a Problem-First Approach” course on Maven — an incredible experience led by Aishwarya Naresh Reganti and Kiriti Badam. This was truly a zero-to-one course — by the end, I could design and build an agentic system powered by multiple RAG pipelines, optimizing for cost, latency, and accuracy, while keeping answers relevant, safe, and empathetic for patients. 💡 For the capstone project, our team built Preppy — a patient assistant that helps people prepare for surgeries and complex medical procedures. Preppy tackles preoperative anxiety, reduces clinician time spent on repetitive questions, and evolved from a basic chatbot into a multi-agent, emotionally aware healthcare companion. 🏆 Our project, Preppy: A Patient Assistant to Help Prepare for Complex Procedures & Surgery, won 1st place among 50+ teams! Huge thanks to my amazing teammates — Abhinav Kolhe, Bindu Koduru, Joshua Pelletier, Kirstin Hartos Maurer, Rohit Maheshwari — for making this journey so collaborative and inspiring. 🙏 Special thanks to Kate Clough for funding and supporting this course, and to Bharath Vemula for recommending it in the first place — grateful you did! 🙌 And of course, immense gratitude to Aish & Kiriti for crafting a program that bridges clarity, creativity, and real-world application. Here’s to building AI that doesn’t just automate — but empathizes, reasons, and truly helps people

Nov 13, 2025

Nguyet McSherry

Just wrapped up the Building Agentic AI Applications course, an inspiring and energizing experience. The most impressive part wasn’t just the frameworks or hands-on materials (though I’ll definitely rewatch them for a deeper understanding), but how current and real the weekly guest conversations were. Every session felt plugged into what’s actually happening in AI right now. Beyond the content, the real value was the community - thoughtful, curious builders exploring how to bring agentic ideas to life. That network is something I’m grateful for. Thank to Kiriti Badam Aishwarya Naresh Reganti. Excited to apply these learnings as I continue designing agentic experiences within my own product workflows. lnkd.in/gYdPN9My

Nov 13, 2025

Varun Goel

Helping the World Work Smarter, One Product at a Time | Senior Product Manager at Mitratech, Ex-Moody’s | PMP, CSPO, AI/ML Certified

Over the past couple of years, I’ve invested fully into the AI wave after watching it reshape how we live and work. It started as certifications and courses and moved on to watching talks, reading articles, and experimenting with the latest features. Basically trying to keep up with a constantly changing scene that has new announcements every day. Between all of this, there was always one gap. We have all this incredible technology at hand, but most mainstream resources do not provide a practical framework to actually implement it at an enterprise scale. And that framework can’t be one and done either. It needs to adapt to how the scene is shifting, while still staying grounded in first principles. My search to fill this gap, combined with Mitratech’s push toward building agentic AI solutions, brought me to "Building Agentic AI Applications with a Problem-First Approach" by Aishwarya Naresh Reganti and Kiriti Badam. It honestly is one of a kind. It takes the academic rigor of a college class and mixes it with the hands-on approach of a real-world workshop. The first thing it tells you is not to chase the most hyped (aka expensive) approach, but to focus on the problem. Unlike a lot of online AI resources, this course emphasizes enterprise use cases, with decision levers like cost and latency optimization, safety and regulatory guardrails, evaluation best practices, and human in the loop design. Speaking of human in the loop, this is one of the few courses where you actually have the instructors available on Slack and through weekly office hours. And to top it off, there’s the capstone element at the end. As with any learning exercise, applying something as soon as you’ve learned it is the best way to lock it in. I was able to put weeks of knowledge into a practical solution, starting from a simple workflow and RAG setup and then exploring orchestration and agentic patterns where they made sense. And as always, I had a lot of fun with the presentation (I’ll share the recording in a separate post!). Finally, even more than the spectrum of enterprise AI (from meta-prompting to context engineering, RAG to fine-tuning, MCP to A2A, and planning to optimization), my main takeaway was a mindset shift. I now have a framework that includes reading papers on AI (kickstarting the habit of reading research papers after a decade is an achievement in itself), using an enterprise lens while building AI solutions, and telling substance from noise to actually build and scale agentic AI. I highly recommend this to anyone looking for something similar. If you have questions about the course, feel free to reach out. The next cohort starts in late January 2026:

Over the past couple of years, I’ve invested fully into the AI wave after watching it reshape how we live and work. It started as certifications and courses and moved on to watching talks, reading articles, and experimenting with the latest features. Basically trying to keep up with a constantly changing scene that has new announcements every day. Between all of this, there was always one gap. We have all this incredible technology at hand, but most mainstream resources do not provide a practical framework to actually implement it at an enterprise scale. And that framework can’t be one and done either. It needs to adapt to how the scene is shifting, while still staying grounded in first principles. My search to fill this gap, combined with Mitratech’s push toward building agentic AI solutions, brought me to "Building Agentic AI Applications with a Problem-First Approach" by Aishwarya Naresh Reganti and Kiriti Badam. It honestly is one of a kind. It takes the academic rigor of a college class and mixes it with the hands-on approach of a real-world workshop. The first thing it tells you is not to chase the most hyped (aka expensive) approach, but to focus on the problem. Unlike a lot of online AI resources, this course emphasizes enterprise use cases, with decision levers like cost and latency optimization, safety and regulatory guardrails, evaluation best practices, and human in the loop design. Speaking of human in the loop, this is one of the few courses where you actually have the instructors available on Slack and through weekly office hours. And to top it off, there’s the capstone element at the end. As with any learning exercise, applying something as soon as you’ve learned it is the best way to lock it in. I was able to put weeks of knowledge into a practical solution, starting from a simple workflow and RAG setup and then exploring orchestration and agentic patterns where they made sense. And as always, I had a lot of fun with the presentation (I’ll share the recording in a separate post!). Finally, even more than the spectrum of enterprise AI (from meta-prompting to context engineering, RAG to fine-tuning, MCP to A2A, and planning to optimization), my main takeaway was a mindset shift. I now have a framework that includes reading papers on AI (kickstarting the habit of reading research papers after a decade is an achievement in itself), using an enterprise lens while building AI solutions, and telling substance from noise to actually build and scale agentic AI. I highly recommend this to anyone looking for something similar. If you have questions about the course, feel free to reach out. The next cohort starts in late January 2026:

Nov 13, 2025

SP

Sujana Pinjala

Expert Engineer | Lead Software Engineer at JPMorgan Chase & Co.

🌟 My First Ever AI Journey — From Curiosity to Confidence When I signed up for my first-ever AI course, I wasn’t sure what to expect. Coming from a non-AI/ML background, stepping into the world of AI felt like learning a whole new language — fascinating, challenging, and full of possibility. But that changed. Over the weeks, Aishwarya Naresh Reganti and Kiriti Badam built something far more powerful than just technical understanding — they built confidence. I went from being curious about AI to actually building AI solutions. Concepts I had only heard in passing suddenly started clicking. Before I knew it, I was applying them in my daily assignments. And then came the capstone project — what a ride! It reminded me of my E2 days — the late-night conversations, multiple iterations, the excitement, and that beautiful sense of progress. Thank you to my excellent team Chandrasekhar Muthyala , Pritish Dhavan , Usha Kasiraman , Shanti Chilukuri , Sreeni Kanna , Juan Carlos Benayas Arroyo But the moment that will stay with me forever was during the final lecture, when Aishwarya Naresh Reganti said: “You belong here! Don’t let anyone tell you otherwise.” That line hit deep — it made me teary-eyed and proud of how far I’d come. A heartfelt thank you to Aishwarya Naresh Reganti , Kiriti Badam , Sahana Venkatesh , Ashwin Naidu — for not just teaching AI, but for instilling a mindset that starts with the problem first. This is not the end — it’s just the beginning. A beginning of a refreshed mindset and a new way of looking at challenges at work. Now, I see possibilities where I once saw limitations. Grateful for this journey, and excited for what’s next. 🙌 Thank you Sai Haritha Chalikonda for introducing and talking to me about your experience with the course ! #AI #LearningJourney #ContinuousLearning #ProblemSolving #Mindset #Growth

Nov 13, 2025

SS

Santosh Sidnal

Excited to share that this week I have completed the “Building Agentic AI Applications with a Problem-First Approach” course by Aishwarya Naresh Reganti and Kiriti Badam. This course is designed for professionals seeking to up-skill their AI skills with balancing their day jobs. It provides a robust foundation and profound, pertinent knowledge in this rapidly evolving technology domain. During the course enough emphasis is given to problem-first approach on solving business problems and how does iterative design is going to to be helpful in building application to be more deterministic. The learning structure is a distinctive feature: weekly content is intelligently categorized into Core, Grow, and Build modules. This innovative tagging enabled to prioritize essential lessons (core & build) and allowed to catch up on advanced concepts at my own pace, ensuring I remained on track. This flexibility is truly what makes it viable for high-demand careers. As part of the course, there were live sessions with various AI industry experts provide invaluable opportunities to interact, network, and gain insights into real-world applications of the knowledge you have acquired. What truly future-proofs this investment is the lifetime access to all future cohort content and updates, ensuring you consistently stay up to date with rapidly changing AI landscape. Course Link: lnkd.in/gur_WDNe

Nov 13, 2025

A

Arjun K.

Learner | Leader | Data Ninja

I’m excited to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach. This course provided a rigorous and hands-on understanding of RAG systems and Agentic AI workflows. While I’ve completed other programs on these topics, the cohort-based structure and problem-first methodology truly differentiate this one as the format forces you to apply concepts immediately, not just learn theory. The curriculum goes beyond surface-level RAG mechanics and dives into real-world design patterns, evaluation frameworks, architectural trade-offs, and production considerations. You build intuition around when and why to choose specific approaches (vs. just how), including cost, latency, complexity, and maintenance trade-offs across orchestration frameworks, vector databases, LLM runtimes, and toolchains. Every week you’re iteratively shipping assignments that simulate real enterprise use cases from retrieval optimization and agent flows to tool calling, orchestration, and observability. The instructors and TAs are experienced practitioners, and the guidance you receive on debugging, evaluation, and scaling patterns is extremely valuable. If you're serious about building reliable, cost-efficient, and production-grade Agentic AI systems whether you're coming from a data engineering, MLOps, software engineering or Product management background: this program will accelerate your practical understanding and execution. Highly recommended! Thank you Aishwarya Naresh Reganti Kiriti Badam.

Nov 13, 2025

Vidhya Sriram

AI Transformation & Implementation | AI Product-Market Fit | Agentic AI Builder | Business Model Innovation

"Building Agentic AI Applications with a Problem-First Approach" - An outlier course + Discount code(EARLY25) As someone who got a refund in week 3 from one of the Maven AI courses, this one actually delivered on its promises and went above and beyond. There are a lot of good courses with AI content out there, but Aish and Kiriti create an inclusive and highly responsive support system that actually makes us opt-in—meaning it makes us voluntarily demand more out of ourselves and feel taller at the end of every assignment. This problem-focused AI course stays true to its commitment: problem-focused thinking before jumping to solutions and implementations. But they also recognize our need to be aware of and apply various technologies. No shortchanging there. Here's what sets them apart: They help us develop a rubric for decision-making—the pecking order of when to use/apply what, and how to always weigh the dimensions of cost, effort, performance, and latency. This kind of framework is about building our judgment about tradeoffs—it is more than becoming competent but enables us to differentiate ourselves in a saturated job market. This course is more than developing competency about agents—it cultivates a sense of belonging and a quiet confidence! Aishwarya Naresh Reganti and Kiriti Badam are the ultimate power couple who empower the rest of us!

Nov 13, 2025

Aiza Hasib

AI Product Manager @ Cisco | MS UX Design | BS Computer Science | Google WTM & GHC Scholar

Building learning products, I can truly appreciate the thought Aishwarya Naresh Reganti and Kiriti Badam have put into designing their course; Building Agentic AI Applications with a Problem-First Approach and building such a strong community! It was very refreshing to learn from their years of experience in the space on what is noise and what matters. The course really helped learn how to cut through all the hype and deeply understand the fundamentals. Being able to apply the concepts through hands on assignments and learning how to iteratively architect and build an AI product was the most valuable part of the course. I really liked their approach of breaking the lectures into core, grow, build modules and being able to choose between Langflow or Langchain for implementation. Even after completing, there is still so much to go back to and I am looking forward to going through all the additional material with the Langchain path and diving deep into Evals with Opik. Also grateful for my capstone team members that I got to learn so much from through all our evening discussions Sampath Putta Lakshmi Devesh Kumar Miguel Vilá Sonali Parthasarathy Veerakumar Yenamadala Thank You to Danielle Hatt and Ashley Roach for being supportive of such growth and learning opportunities!

Nov 13, 2025

A

Abhi🔹 M.

PMP | AI Practitioner | Scaling Teams & Products from 0 to 1 | Website: amoprojects.com

>“AI is just an API call”. On the surface, it might seem that way. But that API call is just the 10% you see: the tip of the iceberg. The other 90%, 𝐭𝐡𝐞 𝐩𝐚𝐫𝐭 𝐮𝐧𝐝𝐞𝐫 𝐭𝐡𝐞 𝐰𝐚𝐭𝐞𝐫, is what makes (or breaks) a reliable, secure, and valuable product. That's why I spent my recent weekends 🧑‍💻 in an intensive programme on AI system design with Aishwarya Naresh Reganti and Kiriti Badam, learning from their "skin in the game" experience at Amazon and OpenAI. We covered topics such as: 🏛️ 𝐒𝐲𝐬𝐭𝐞𝐦 𝐝𝐞𝐬𝐢𝐠𝐧 (Knowing when to use which architecture – RAG vs Multi-Agents) 🎯 𝐄𝐯𝐚𝐥𝐬 (How do you even know the AI is accurate?) ⚖️ 𝐆𝐮𝐚𝐫𝐝𝐫𝐚𝐢𝐥𝐬 (How do you manage cost, security, and risk?) My biggest takeaway? Building enterprise-grade AI solutions is both an organisational and team effort, not a solo one. 🤝 A huge thanks to my capstone team as well (you know who you are) for making a unique project on government RFPs so much more impactful. What's the biggest barrier you've found moving AI initiatives from a POC to production? Comment below! 👇 #DigitalTransformation #EnterpriseTech #SystemDesign #Innovation #Product #AI

Nov 12, 2025

Sampath Putta

Senior Test Automation Systems Engineer at Rivian Volkswagen Tech | Building Automation Systems for a Greener Future | RF/Wireless Validation Specialist

I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven! This course has taught me so much about inner workings of AI space and learning the fundamentals through “learning by doing” approach of instructors of the course : Kiriti Badam Aishwarya Naresh Reganti is just so insightful and useful. Thank you!! They have built a whole new vibe through the course and slack group.. really excited to have joined this course and I highly recommend to anyone interested to get themselves updated on AI concepts. Want to see what we have built in this course? Join us in today evening session (6th team in the evening session). It’s free. There is morning demo sessions by fellow cohort members on lot of cool ideas. : lnkd.in/gpPhyQTW Thank you so much to the team who joined me on capstone project that we would be demoing this evening. Cheers to many evening sessions and discussions in bringing something that we all could utilize .. Fantastic and passionate team members: Aiza Hasib Lakshmi Devesh Kumar Miguel Vilá Sonali Parthasarathy Veerakumar Yenamadala

Nov 12, 2025

Nitya Thiyagarajan

Product Leader @Docusign | Ex-Amazon

Just wrapped up Aishwarya Naresh Reganti and Kiriti Badam’s 'Problem First AI course', a hands-on learning experience that pushed me to think deeply about how AI can truly solve real world problems. When I started, I was honestly overwhelmed by the sheer volume of AI content out there. Aishwarya and Kiriti cut through the noise, grounding everything in problem first product thinking rather than jumping straight into tools. We explored how to approach AI product development from first principles, balancing what’s feasible and valuable. We went from understanding core concepts to building, and ultimately to selecting a real world enterprise problem, thinking through how to iterate on it and where AI can be most effectively applied. From cost and latency tradeoffs to evaluations and guardrails, the course brought a lot of clarity on where GenAI and ML actually add value in an enterprise setting. One big realization, learning in AI will only continue. The field is evolving so fast that curiosity and experimentation are the only sustainable edges. Thank you to Aish and Kiriti, two passionate educators who created such an incredible learning experience and inspired so many of us to approach AI the right way.

Nov 12, 2025

AV

Ashwini Kalkunte Vikram

Product Leader 🧭 On a Purposeful Break | Reskilling, Reflecting & Giving Back Through Storytelling

Aishwarya Naresh Reganti and Kiriti Badam Excited to learn from you, with this huge brilliant cohort "Building Agentic AI Applications with a Problem-First Approach" lnkd.in/g3xx9Di3 #AI #AgenticAI

Nov 12, 2025

Abhishek Kumar Singh

Exploring what’s next at the intersection of Engineering, Security, and AI.

Just completed the “Building Agentic AI Applications with a Problem-First Approach” course — and it was truly worth the time! Course Design: The content is really well-structured — it helps you focus on the right areas step by step. You never feel lost, even though there’s a lot of depth packed into the material. Trainers: Aishwarya and Kiriti are exceptional. They bring strong practical experience and share how real-world companies are building agentic systems, which makes the sessions insightful and relatable. Hands-on Experience: What I liked most — it’s not just theory. You actually build working AI applications yourself, which gives a solid hands-on feel of what goes behind real agentic AI systems. A great learning experience overall!

Nov 12, 2025

Poorna Ramakrishnan

Accenture AI | Analytics Manager

Ever found yourself aware of the tools, but unsure when and how to apply them effectively? The "Building Agentic Applications using a Problem-First Approach" course by Aishwarya Naresh Reganti and Kiriti Badam has been incredibly valuable in learning a structured, step-by-step way to solve real AI problems. It covers the full spectrum — from prompt design and RAG to multi-agent systems — while emphasizing problem-first thinking over tool-first execution. What truly stands out is the focus on evaluation frameworks and optimization strategies, an area often overlooked yet critical for building scalable and reliable GenAI solutions. Huge thanks to the instructors for an insightful learning journey, and to Sahana Venkatesh and Ashwin Naidu for always being super responsive and supportive. I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven!

Nov 12, 2025

Bhusan C.

AI & Data Science Leader | Generative AI & LLM Expert | RAG Architect | Agentic & Voice AI Systems | Built AI Divisions from Scratch

Just finished the course “Building Agentic AI Applications with a Problem-First Approach” by Aishwarya Naresh Reganti and Kiriti Badam. This course offered a truly structured and practical path to understanding how to design Agentic AI systems - starting from identifying the right problem, designing iteratively, and keeping evaluation and reliability at the core of every stage. There’s so much noise online regarding AI agents, and it’s so easy to get lost. But with this course and the accompanying materials (on Maven and GitHub), I didn’t have to look anywhere else - everything I needed was right there in a credible, well-organised form. I’m really glad I decided to take this course despite everything else going on - it was absolutely worth it. Course details:

Nov 12, 2025

Priyank Khanna

Just wrapped up something big — Building #Agentic AI Applications with a Problem-First Approach! In an era where everyone’s chasing the latest AI buzzword, this certification pushed us to do the opposite — to start from #real business problems, not shiny tools. Over the past weeks, I learned to #design and build agentic AI systems that go beyond demos — systems that solve measurable, high-impact business challenges. 🔍 Key takeaways: 1. #Framing problems through outcomes, not features or model names. 2. Using RAG, agents, #evals, and MCP — but always in #context 3. Building multi-agent #architectures that reason, retrieve, and reflect. 4. Applying evaluation and #guardrails to make AI systems reliable, not just impressive. 5. Balancing tradeoffs between #accuracy, #latency, and #cost in production-ready designs. The highlight? Collaborating with brilliant peers to design an end-to-end agentic AI solution — Will be having live demos of what we build. Check out in the following links: lnkd.in/gHaXcQ8n lnkd.in/gvBVVx_G This wasn’t about “playing with #prompts. ” It was about learning to think in systems, question assumptions, and bridge the gap between AI capability and business value. Big thanks to the mentors Aishwarya Naresh Reganti and Kiriti Badam, and community who made this experience transformative. 🙏 If you’re building or scaling AI systems, I’d love to connect and exchange ideas on real-world agentic design — beyond the hype. 💡 #AI #AgenticAI #RAG #MultiAgentSystems #ArtificialIntelligence #Learning #Innovation #ProblemSolving #MCP #AIEngineering #BusinessImpact

Nov 12, 2025

Alejandro Ayala-Hurtado

Senior Software Engineer, Senior Backend Engineer, Senior Product Engineer, Engineering Manager, AI Engineer | Python, Golang, Kafka, LangChain, LangGraph, FastAPI, Typescript, JavaScript, React, Node.js, GCP | Waltzer

I just completed Aishwarya Reganti and Kiriti Badam’s course on Building Agentic AI Applications with a Problem-First Approach. As someone still fairly new to the field, this was an excellent introduction to modern Agentic AI systems—from prompt engineering, workflow agents, evals, tool calling, MCP, and RAG to autonomous and multi-agent systems. What stood out most was the "problem-first" mindset—learning how to apply AI thoughtfully to real enterprise challenges. The course provided a structured framework for identifying valuable problems and building practical AI solutions around them. We also had guest lectures from industry practitioners, which made it clear how these concepts are being applied in real organizations today. Another thing I appreciated: the course offered different tracks depending on your background and technical depth. I chose the most technical path, but the material was accessible and relevant to learners from all backgrounds. Overall, I really enjoyed this experience and came away with a much clearer understanding of how to build and reason about AI systems that actually solve business problems. Huge thanks to Aishwarya Naresh Reganti, Kiriti Badam, Sahana Venkatesh, and Ashwin Naidu for creating such a thoughtful and inspiring course—and supporting us throughout the journey!

Nov 12, 2025

Liz Filardi

Business Ops and Growth Consultant | CS leader at Hubspot

I can't believe I just built a deep research multi-agent system, but I know Aish and Kiriti would be shaking their heads if that was all I took away from their course! Honestly, I went in hoping to develop AI literacy and informed opinions about how/when to leverage AI. Taking a course to learn how to build agents alongside engineers seemed like the scariest way to go about that, and somehow it was exactly what I needed. I realized I had grown a ton when a friend asked for my take on an AI tool that he was considering purchasing for his business. I had a lot to say! I could tell him that the tool was built by custom fine-tuning on top of an existing LLM, though we don't know which one. I told him that I thought the functionality was very limited for the price tag, seeing as the only way to use it was through a chat interface, so while you could ask it to build agents for you, you would never be able to efficiently troubleshoot and iterate, build evals and optimizations, you can't see what's going on under the hood with the architecture, and you can't leverage it as an LLM in a workflow tool like n8n, so that also limits the capabilities... the conversation went on. The point was less about the tool, and more about his use case. Why do you need a fine-tuned LLM? What are you trying to achieve? That's the problem-first approach. I know now that the industry is quite open about innovation with papers coming out seemingly daily and documentation galore, and I know where to access that stuff and why it matters in my day to day. I know what AI slop looks like and how to get past the hype to find the best value for my skill and knowledge level so that I can build things efficiently. I have many strong ideas about where I need to augment my work with AI, and where I need to use my brain instead. I know how to start small and simple and build from there, what is worth hours and hours of my time (evals, guardrails, optimizations) and what is not (trying to make something robust using a tool with very low interoperability and low transparency). This course has been a launch pad for my learning and growth and has gotten me excited about my career again. It is in no small part owed to the instructors' passion for the space, and tireless support of our learning process. They found some way to beam that passion into all of us in the cohort in such a way that we've now got some 75 alums in a study group slack channel where we plan to continue our lessons and experiments. Most of all, I am deeply grateful for how they have built a diverse community, recognizing that community is where growth happens, and that we all benefit from a space where we can do that out in the open. Thanks a mil, Aishwarya Naresh Reganti, Kiriti Badam, Sahana Venkatesh, Ashwin Naidu, and all the amazing people in my cohort! P.S. I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven!

I can't believe I just built a deep research multi-agent system, but I know Aish and Kiriti would be shaking their heads if that was all I took away from their course! Honestly, I went in hoping to develop AI literacy and informed opinions about how/when to leverage AI. Taking a course to learn how to build agents alongside engineers seemed like the scariest way to go about that, and somehow it was exactly what I needed. I realized I had grown a ton when a friend asked for my take on an AI tool that he was considering purchasing for his business. I had a lot to say! I could tell him that the tool was built by custom fine-tuning on top of an existing LLM, though we don't know which one. I told him that I thought the functionality was very limited for the price tag, seeing as the only way to use it was through a chat interface, so while you could ask it to build agents for you, you would never be able to efficiently troubleshoot and iterate, build evals and optimizations, you can't see what's going on under the hood with the architecture, and you can't leverage it as an LLM in a workflow tool like n8n, so that also limits the capabilities... the conversation went on. The point was less about the tool, and more about his use case. Why do you need a fine-tuned LLM? What are you trying to achieve? That's the problem-first approach. I know now that the industry is quite open about innovation with papers coming out seemingly daily and documentation galore, and I know where to access that stuff and why it matters in my day to day. I know what AI slop looks like and how to get past the hype to find the best value for my skill and knowledge level so that I can build things efficiently. I have many strong ideas about where I need to augment my work with AI, and where I need to use my brain instead. I know how to start small and simple and build from there, what is worth hours and hours of my time (evals, guardrails, optimizations) and what is not (trying to make something robust using a tool with very low interoperability and low transparency). This course has been a launch pad for my learning and growth and has gotten me excited about my career again. It is in no small part owed to the instructors' passion for the space, and tireless support of our learning process. They found some way to beam that passion into all of us in the cohort in such a way that we've now got some 75 alums in a study group slack channel where we plan to continue our lessons and experiments. Most of all, I am deeply grateful for how they have built a diverse community, recognizing that community is where growth happens, and that we all benefit from a space where we can do that out in the open. Thanks a mil, Aishwarya Naresh Reganti, Kiriti Badam, Sahana Venkatesh, Ashwin Naidu, and all the amazing people in my cohort! P.S. I’m happy to share that I’ve obtained a new certification: Building Agentic AI Applications with a Problem-First Approach from Maven!

Nov 12, 2025

VS

Vikas Singh

Software Development Manager at Amazon

Just wrapped up an incredible learning journey with instructors Aishwarya and Kiriti, who delivered a masterclass in AI fundamentals. What sets this course apart is its focus on core concepts rather than getting caught up in the endless cycle of tools and frameworks—a refreshing approach that cuts through the AI noise. Key takeaways that strengthened my AI foundation: • Prompt Engineering techniques • RAG (Retrieve-Augment-Generate) implementation • ReAct frameworks • Planning Agents architecture • Agentic RAG systems The course strikes the perfect balance between theory and practice, with hands-on assignments that reinforce learning while remaining accessible regardless of your coding background. The problem-first methodology ensures you understand the "why" before diving into the "how"—invaluable for building enterprise-level AI solutions. Feeling genuinely "AI-ducated" and equipped with transferable skills that adapt to any tool or framework. Huge thanks to Aishwarya Naresh Reganti and Kiriti Badam for creating such a comprehensive and well-structured program! 🚀 #AI #AgenticAI #Maven #ProblemFirstApproach

Nov 12, 2025

Wall of love for Building Agentic AI Applications with a Problem-First Approach