_

I completed my M.Tech in Artificial Intelligence at IIT Jodhpur in 2025, and I've spent the time since building the kind of systems that used to only live in research papers, agents that use tools, retrieval pipelines that actually cite their sources, models fine-tuned until they stop guessing.

At Turing, most of my work has sat at the intersection of data science and applied GenAI: designing evaluation frameworks for LLMs, authoring SFT datasets that measurably improved tool-use and multi-step reasoning on a 15B-parameter model, and building the internal tooling that made all of it repeatable rather than one-off.

What actually drives me is turning ambiguity into something testable. Most interesting ML problems aren't modeling problems, they're the fuzzy human ones underneath: what "correct" means here, how you measure it, how you get two teams to agree on it. I like working exactly there, between the whiteboard and the notebook, and shipping something that holds up.