LevelUp Labs ads

We kicked off yet another cohort of our course "Building Agentic AI Applications with a Problem-First Approach" today! At the end of every session, we ask our students: What’s the most valuable thing you learned or unlearned today? Our students shared so many thoughtful takeaways today, and it’s incredibly rewarding to see how much one session can shift their thinking.Here’s a small sample of what they learned and unlearned (image below) Unlearning is a big part of the course. With so much noise around AI, learning also means questioning assumptions and letting go of things we’ve picked up along the way. We’re still accepting registrations until tomorrow. If you’ve been thinking about joining, we’d love to have you be part of the party! Sign up here: https://lnkd.in/e-pYU4jq (use code PF25 for a 25% off)

"Don’t build AI prototypes just for the heck of it. Build thoughtfully, starting with the problem you’re trying to solve." has never been more true! Kiriti Badam & I have been saying this for the past two years, and I think 2026 has made it even more important. There’s so much more we can build now, and it’s easy to get pulled into trying every new tool without being clear on why you’re using it. So pumped that we are bringing those lessons into our last Problem First AI cohort of 2026 (starting tomorrow)! We’ve updated a big part of the syllabus, with more on multimodal AI, designing deep agents and building harnesses. These updates are grounded in our own experience building these systems. We’ll work through when these approaches are useful, what problems they solve and how to build with them thoughtfully. Teaching this course has given me so much joy over the past two years. We’re so pumped to get back into it and welcome our newest members into the community. It’s that time again! Link: https://lnkd.in/e-pYU4jq

I recently completed the 5-week Maven course on “Building Agentic AI Applications with a Problem-First Approach”. - https://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.

🥳 We’re ending 2026 with a bang. Our Full Stack AI Bundle gives you a clear path to build the skills, portfolio, and confidence you need to pivot into AI by the end of 2026. The bundle includes: - Our flagship AI cohort, where we take a problem first approach to help you go from zero to one on AI foundations and build your own enterprise AI capstone. Our Oct Curriculum includes the newest updates, including modules on voice AI, multimodal AI, AI inferencing 101, and more! - Our advanced self paced courses on Harness Engineering and AI Evals, designed to help you get much more confident with the skills becoming increasingly important for AI engineers and builders. - Partner build sessions throughout the holidays and during the cohort, featuring some of the latest tools and companies in AI. $800 to $1,000 in credits to support your learning and building. It’s hard to ask for a better package if your goal is to transition into AI. You’ll learn the foundations, build real projects, strengthen your resume and portfolio, and come out much more confident about actually building with AI! You can enroll directly through our course page: https://lnkd.in/e-pYU4jq Use code "FSEARLY" when you enroll, and we’ll apply the full bundle and give you access to the additional courses.

I have spent 6 months learning Generative AI on my own before this course, so I was not starting from zero. As part of my learning, created my own Model Context Protocol and created couple of AI Agents as well. All this was done before the course. I joined the Maven course "Building Agentic AI Applications with a Problem-First Approach" after talking to the previous cohort members as they were mighty impressed with the course. Now that I have joined the course and almost completed it, I can 100% say that this has been totally worth it. After this course, I started thinking in first principles, and now have a solid foundation of Prompt Engineering, RAG, Memory, Tool calling, AI Agents, and MCP. I have also started reading research papers related to Gen AI. All thanks to Aishwarya Naresh Reganti and Kiriti Badam. On top of all that, there is an amazing network of alumni from the cohort and super-awesome guest speakers Pritika Mehta and Andre Kerr from the Industry. Icing on the cake is Chai & AI which is super hot with it's hot takes. It was great to learn from Francisco D'Souza and Jaya Gupta. Course URL: https://lnkd.in/db3sP4r5 Feel free to reach out to me if you have any questions choosing this course. Even though this is a GenAI course, this feedback and all it's grammatical mistakes were typed by my own fingers :) #ProblemFirstAI #LLM #GenAI

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 - https://bit.ly/4j3FQ7p