About

I work at IBM Research on time-series foundation models — how they're trained, how you tell whether the data they learn from is actually pulling its weight, and what it takes for them to hold up on domains they were never shown. Alongside that I build agentic systems that reason over streaming data: things that retrieve context, use tools, and keep enough memory to stay coherent as new observations arrive.

I studied at IIT Bombay, where I did a dual degree in Electrical Engineering with a specialization in communication and signal processing, and a minor in Data Science & AI. That background — signals first, learning second — still shapes how I think about temporal data: as something with structure worth respecting rather than a flat sequence of numbers.

My M.Tech thesis was on uncertainty modeling for open-domain generalization — how a model should express doubt when it's asked to work on domains, and label spaces, it has never encountered. The through-line across most of what I do is the same: getting systems to know the edges of what they know.