An independent laboratory, not a company.

“Labs” here means what it means in a university corridor or a garage: a place where things get taken apart, tried, measured and written up. It does not mean a company. There is no team, no service, no roadmap sold to anyone.

The work is deliberately not restricted to one discipline. Sometimes that means building an AI system. Sometimes it means investigating a scientific question. Sometimes it means reading papers until a problem stops being confusing. And sometimes it means building something purely because the problem would not leave me alone.

What holds it together is not a subject but a method: build the thing from scratch, because understanding how something works usually requires having built it. A search engine is easy to describe and hard to write. The gap between those two is where the interesting part lives.

How this site works

This is an archive rather than a portfolio. Entries record what a piece of work is, what state it is genuinely in, and what happened to it — including the parts that failed, were rejected, or turned out to be the wrong idea. A project that was refused is more useful to read about than one described only at its best moment.

Several sections are currently empty. They stay empty until there is something real to put in them. The site is meant to fill up over time rather than launch complete.

What is happening now

The active build is Anvesha — an open-source, modular search platform being written from scratch and in public, as a deep dive into search engines, information retrieval and distributed systems. It is in its architecture and design phase.

Before that, PranSaathi — an application for managing patient queues in government hospitals, with ABDM integration and district-level outbreak prediction. It was presented to the ABDM Joint Director of Jharkhand and rejected on unclear compliance grounds, which was itself the most instructive part.