story

I'm an applied AI engineer at Clueso (YC W23) in Bengaluru. We make product videos and docs with AI, increasingly end to end: agents that study your product, script it, record it in a real browser, and re-render when the code changes.

Most of what I know comes from shipping agents to production and building the infra underneath them. The rest comes from small tools nobody asked for.

Straight out of college I joined Omni RPA as the founding AI engineer and got handed the fun kind of problem: build the entire AI backend for a cloud automation platform, from nothing.

That turned into a multi-agent DAG orchestrator with eight agent types, an MCP server built while the protocol was five weeks old, a knowledge-graph and RAG stack that did most of its entity extraction locally, semantic memory on pgvector, a fine-tuned model for constraint extraction, and the OTel plumbing to know what all of it cost. I was also primary on-call, which is a very effective way to learn which of your ideas were bad.

B.Tech at VIT Chennai, AI & ML specialization. The part that actually mattered: leading a team that built an autonomous medicine-delivery robot for hospitals. Raspberry Pi, obstacle detection, path following, RFID room identification. It ended up published in Springer LNNS. Somewhere in between I spent a research winter fine-tuning YOLOv8 to tell ripe crops from unripe ones, and it shipped to production at Digital University Kerala.

I ship small tools compulsively. A docker clone for macOS started as an overnight bet and ended up on Homebrew. Most of what I make lives in the terminal, because that's where I live too.

When I'm not doing that: sending indie devs unreasonably detailed product feedback on X, and watching football. Lamine Yamal will win everything, you heard it here.