I build AI systems and the products that carry them.

I love building things, breaking them on purpose to see how they work, and picking up whatever's necessary to make the next idea real. That's mostly how I ended up specializing in AI on top of full-stack: every interesting feature I got asked to build turned out to be a full-stack problem with a retrieval system hiding inside it. I write Go and Python about as often as TypeScript.

Most recently, at Satven, I built the RAG pipeline that checks automotive design documents for compliance — the kind of review that used to eat an engineer's whole afternoon. Automating it cut that process time by about 30%. Before that, at Savemom, I built the landing page, and contributed to their superadmin portal.

Outside of work I'm usually mid-experiment on something I didn't strictly need to build. Like,Kortex, that started has a hackathon project has evolved into an ai application that tries to teach a topic by building intuition instead of just handing over answers. BloomIQ started as a way to generate and manage my own college's exam question papers based on Bloom's Taxonomy without the data ever leaving campus. Relay — an n8n-style workflow engine I built after watching n8n itself and wanting to understand how something that well-engineered actually holds together: durable execution, live status, credentials done properly. I learn a stack by rebuilding something that inspires me. Most of it ends up on GitHub, half-documented, and half finished, but the completed ones are here for you to see