About

I started Vedic AI when I was 17. What began as an attempt to make Jyotish software more personal has gradually taken me into AI agents, astronomical calculations, structured knowledge, retrieval systems, and the less visible parts of building a consumer product.

Before it, I worked on SwaraConnects, an earlier project around students, learning, and community. That introduced me to a different kind of problem: having an idea for a social product is one thing; understanding why people participate and come back is another.

With Vedic AI, I became interested in what would happen if an application treated a birth chart as structured information instead of asking a language model to generate an entire reading from a prompt.

Calculating a chart isn't the same as interpreting it. Retrieving a rule isn't the same as knowing whether it applies. And a fluent answer isn't necessarily a well-supported one.

Those questions kept leading me deeper into the engineering.

I'm especially interested in systems where different kinds of work have to fit together: computation, knowledge, retrieval, model decisions, and product experience.

Outside the company, I've experimented with smaller language models and spent time learning more about tokenization, training, evaluation, and limited-compute workflows.

I enjoy understanding how software works beneath the interface. But I've also learned that technical depth alone doesn't make a good product. People notice whether an answer makes sense, whether the app is slow, whether it fails, and whether they have a reason to return.

My main focus is Vedic AI and the deeper consultation experience we're building through Drishti.

I'm working on how the agent selects evidence, uses calculation tools, handles difficult questions, and gives people a useful experience without unnecessary complexity. I'm also learning the business side: distribution, speaking to users, pricing, and finding out which assumptions hold up outside a demo.

There's still a lot I don't know. I'd rather work on those questions than pretend they're settled.