How I went from B.Tech student to AI Researcher at Amrita Vishwa Vidyapeetham, with stops at medical AI research, NASA Space Apps, and publishing along the way.
If you had told me in my first year of B.Tech that I'd be an AI Engineer working on production agentic systems, I wouldn't have believed you. The path from student to engineer was anything but a straight line — and that's what made it valuable.
My first two years of undergrad were about building fundamentals. Not just the coursework — the late nights learning Python, the weekend projects that never shipped, the dozens of unfinished GitHub repos.
What actually mattered:
The turning point was getting involved in medical AI research. I joined a research group working on clinical text analysis and knowledge graphs. It was overwhelming — the gap between coursework and actual research was enormous.
Key lessons from this phase:
Research is 90% reading. Before you can contribute anything new, you need to understand what's already been done. I spent months reading papers before writing a single line of research code.
Write to think. The act of writing a paper forces clarity. Vague ideas become precise when you have to explain them to reviewers. I published my first paper during this period, and the writing process taught me more than the research itself.
Find mentors. The researchers who took time to explain concepts, review my code, and push back on my ideas shaped my thinking more than any course.
The NASA Space Apps Challenge was a pivotal moment. 48 hours to build a solution from scratch. No time for overthinking — just ship.
We built an AI-powered solution that won the Local Impact Award. It wasn't the most technically sophisticated project I'd worked on, but it was the one that taught me:
The transition from research to engineering at Amrita AI brought a new set of lessons:
Production is a different game. In research, it works on your laptop. In production, it works at scale, handles edge cases gracefully, and doesn't go down at 2 AM.
Ship, then iterate. The instinct from research is to perfect things before sharing. In engineering, you ship an MVP, get feedback, and improve. It's uncomfortable at first but dramatically more effective.
Write things down. Documentation, design docs, post-mortems. Writing clarifies thinking for yourself and for your team. It's one of the highest-leverage activities an engineer can do.
Build in public earlier. Every project, every failure, every learning — share it. The opportunities that came through Twitter, LinkedIn, and GitHub dwarfed everything from formal applications.
Research and engineering are complementary, not competing. My research background makes me a better engineer (I think about evaluation, edge cases, and rigor). My engineering experience makes me a better researcher (I think about scalability, reproducibility, and practical impact).
The goal isn't to know everything — it's to know how to learn anything. The specific technologies I learned in 2022 are already outdated. The ability to pick up new tools, read papers, and build quickly is what actually compounds.
People > projects. The collaborations, mentorships, and friendships I've built along the way matter more than any single project or paper. AI is a team sport.
I'm focused on agentic AI systems — building architectures where LLMs don't just answer questions but plan, reason, use tools, and accomplish goals autonomously. It's the most exciting engineering challenge I've encountered.
Beyond that: contributing to open-source AI tooling, publishing more research, and eventually founding something. The journey from student to engineer is just the beginning.
How I went from B.Tech student to AI Researcher at Amrita Vishwa Vidyapeetham, with stops at medical AI research, NASA Space Apps, and publishing along the way.
This article was written by Avishek Rauniyar, an AI Engineer and Researcher specializing in Career.