ML Engineer · Founder · ex-ISRO
I build with AI and ship things that work — but I'm just as curious about people as machines. Ex-ISRO, two-time founder, now at TUM studying how data shapes the society it runs on.
// 001 — the person behind the code
Engineering wasn't the plan. I finished school in the COVID lockdown with no real direction — more drawn to history, politics and human psychology than to code. But I'd kept a notebook of ideas for years, and when I finally picked up programming, I realised I could build them.
That curiosity became YellowShuttl (my first startup, 1000+ students), an apprenticeship building AI at ISRO, and Reparo AI — a company I poured myself into and, eventually, watched close.
Now I'm at TUM for a Master's in Data & Society — a return to the questions about people that interested me first, finally with the engineering to dig into them.
// 002 — how i got here
Engineering wasn't really the plan. This is how I actually got here.
Long before any of this, I kept a notebook of everyday problems and how I'd fix them. I was the kid more curious about people and history than code — until the COVID lockdown, when I picked up programming and realised I could finally build the ideas I'd been collecting for years.
A campus fleet system that 1000+ students used daily across 20 buses, with real-time tracking that cut wait times by 68%. It won Best Social Project at the expo — and ended when the university got cold feet about backing two young founders.
Production AI inside ISRO's air-gapped HPC: an internal coding assistant, a meeting assistant with transcription, and secure RAG tools for scientists. I left with a Token of Appreciation — and a full-time offer I turned down to chase an idea of my own.
The idea I left ISRO for: a multimodal assistant that helps people repair and maintain their own devices. I built the whole platform — voice, image and text, a simulation engine generating guides for hundreds of products. We pitched it, nearly funded it, then closed it over co-founder differences. Still the most I have ever learned from anything.
A deliberate return to the questions that interested me first — how data, people and society shape one another. I didn't pick this Master's for a career move. I picked it because it's the thing I've always wanted to study.
// by the numbers
A few years, more or less — and every one of these is real.
// 003 — things I've built
Some shipped, some closed, all of them taught me something. Open any to read the full story.
// also in the archive — 5
Most of my best ideas start by noticing something everyone else walked right past.
— pretty much how I think
// 004 — what I'm actually good at
How I'd honestly rate myself, if you asked.
Hover a strength to see where it lands.
Learned it on the job at ISRO — trained a computer-vision model for fracture detection, lived inside LLMs, and built the core of Reparo on it. I've also taught it, to fellow interns and a few stubborn friends.
Started with YellowShuttl (1000+ daily users). Since then: internal tools at ISRO, a full-stack event platform for a medical conference, and whatever I happen to be building this month.
Comes naturally. Most of my ideas start by watching ordinary moments and connecting them to problems nobody else is looking at. The intuitive part is genuinely hard to pass on — I've tried.
Learned by pitching my own companies to investors. It earned real interest and a funding offer — which I walked away from on the terms.
Leading small teams taught me this. I try to understand the other side first, then bring them to mine — so things resolve without anyone feeling like the loser.
// the tools I reach for
// 005 — recognition
YellowShuttl · University Project Expo
AI/ML Apprenticeship · ISRO
One-22 · state-level business comp · 2022
Srushti-2022 Ideathon · VTU · BMSCE
Picked for San Francisco — the visa, sadly, was not
Head Organizer · 32 volunteers · 4 days