About
I make software for things people stopped questioning.
I'm Agnay, a builder from Mumbai now working on Tsenta in San Francisco. I like ambitious products, boring infrastructure, and deleting repetitive work.
I graduated magna cum laude in Computer Science from Rose-Hulman in 2026. Along the way I worked across systems, MLOps, medical imaging, and product engineering. The useful throughline was not a particular stack. It was learning how to make software survive contact with real people and real constraints.
Tsenta started with a problem my co-founder Pulkit and I knew too well. Between us, we had applied to more than 3,000 jobs. We were losing weekends to the same fields, the same accounts, and the same uncertainty about what happened after submit.
We built the first version for ourselves in 2025, rebuilt it around trust and control, and joined Y Combinator's Summer 2026 batch. Today I'm focused on agents that can do meaningful work without turning the person using them into a passenger.
That is still my filter. I want problems that I have felt myself, or that someone close enough to text me at midnight has felt. Lived frustration is a better product brief than a trend report.
Away from Tsenta, there is usually an overbuilt home server, an Arch Linux machine, a side project, or a barbell nearby.
Work experience, unredacted
The résumé leaves out the useful part.
Slow software charges everyone.
A movie-ticketing app taught me how tiny bits of product friction become expensive at scale.
AI is only useful after the demo.
I worked on LLM traffic, scoring, and observability so the system could survive contact with actual users.
Give experts their attention back.
I automated terabytes of spinal-imaging work so researchers could spend more time on the surgery problem.
$27/hour. No return offer.
I built systems software, learned a lot, and got no return offer. That plot twist pushed my own job-search problem into Tsenta.
The job search became the product.
Pulkit and I built the tool we needed ourselves, then rebuilt it around transparency, control, and a price normal people can afford.
Operating principles.
- 01
Build for a memory of yourself.
Or build for someone close enough to call when it breaks. The sharpest product judgment starts with a problem you have actually lived.
- 02
Trust is part of the architecture.
If software acts on someone's career, money, or identity, visibility and control are core behavior.
- 03
Useful beats impressive.
A small thing that works every day matters more than a demo built to win one room.