The latest articles on DEV Community by Abdulla Al Noman (@abdulla_al_noman). https://dev.to/abdulla_al_noman https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3644294%2F9fe0da5e-3d8f-469f-a507-e7c4b2f8dc07.png https://dev.to/abdulla_al_noman en Abdulla Al Noman Wed, 03 Dec 2025 21:31:11 +0000 https://dev.to/abdulla_al_noman/my-ai-agents-intensive-journey-learning-building-and-reflecting-1nhh https://dev.to/abdulla_al_noman/my-ai-agents-intensive-journey-learning-building-and-reflecting-1nhh <p><em>This is a submission for the <a href="https://dev.to/challenges/google-kaggle-ai-agents-2025-11-10">Google AI Agents Writing Challenge</a>: Learning Reflections</em></p> <p>Hi Dev Community!<br> I’m <strong>Abdulla Al Noman</strong>, a 4th-year CSE student at BRAC University. I recently completed the <strong>5-Day AI Agents Intensive with Google</strong>, and it was an incredible deep dive into the world of AI agents - autonomous systems that can think, plan, and collaborate.</p> <p>Here’s my reflection on the journey: what I learned, what resonated most, and how I applied it in my <strong>capstone project, ScholarMindAI</strong>.</p> <h2> Key Learnings &amp; Concepts That Resonated </h2> <p><strong>Day 1 – Foundational LLMs &amp; Prompt Engineering</strong><br> I realized that <strong>even the smartest AI needs clear guidance</strong>. Crafting structured prompts, tuning parameters, and evaluating responses are critical to making AI agents reliable.</p> <p><strong>Day 2 – Embeddings &amp; Vector Databases</strong><br> This taught me that <strong>semantic understanding beats keyword search</strong>. Embeddings and vector stores let AI agents retrieve and compare information intelligently, enabling smarter research workflows.</p> <p><strong>Day 3 – Generative AI Agents</strong><br> Here, I learned how to <strong>orchestrate multiple agents together</strong>. Function calling, multi-agent design, and connecting agents to real-world tools highlighted how autonomous systems can tackle complex tasks collaboratively.</p> <p><strong>Day 4 – Domain-Specific LLMs</strong><br> Fine-tuning agents for specialized domains showed me that <strong>context matters</strong>. Domain-adapted agents perform better and can handle complex knowledge like research papers or medical data.</p> <p><strong>Day 5 – MLOps for Generative AI</strong><br> Making AI agents production-ready taught me the importance of <strong>resilience, observability, and memory management</strong>. It’s not just about building capabilities - it’s about making them reliable in the real world.</p> <h2> Capstone Project: ScholarMindAI </h2> <p>To put these concepts into practice, my teammate <strong>Suprava Saha Dibya</strong> and I built <strong>ScholarMindAI</strong>, a <strong>multi-agent academic research assistant</strong>.</p> <p>It can:</p> <ul> <li>Search and summarize papers automatically</li> <li>Compare methodologies and highlight insights</li> <li>Generate structured literature reviews</li> <li>Manage citations in APA, MLA, Chicago, or Harvard</li> <li>Build dynamic knowledge graphs connecting concepts, authors, and methods</li> <li>Predict research impact and track project progress</li> <li>Support collaboration with shared dashboards and handoff documents</li> </ul> <p>This project gave me hands-on experience orchestrating agents, handling memory and context, and designing workflows that <strong>actually help humans work smarter</strong>.</p> <h2> Reflections &amp; Growth </h2> <ol> <li>AI agents are <strong>team players</strong>, not just tools - they amplify human curiosity.</li> <li>Memory and context are essential for agents to handle complex, multi-step tasks.</li> <li>Hands-on labs and the capstone made abstract concepts tangible.</li> <li>Production readiness - logging, error handling, resilience - is just as important as intelligence.</li> </ol> <p>The course changed my perspective: I now see AI agents not just as code, but as <strong>autonomous collaborators capable of meaningful work</strong>.</p> <h2> Thoughts </h2> <p>The AI Agents Intensive taught me that AI doesn’t replace curiosity - it <strong>enhances it</strong>. ScholarMindAI is my first step toward building agents that <strong>free researchers to think, explore, and innovate</strong>.</p> <p>I’m excited to continue exploring this space and connecting with others passionate about AI agents.</p> <p><strong>Thanks for reading!</strong> </p> <p>— <strong>Abdulla Al Noman</strong></p> googleaichallenge ai agents devchallenge