Nihal Jain

These days I’m most excited about efficient algorithms for online learning and reasoning that allow machines to adapt to changing environments.

A selection of my current and past work is below. The full list is available on Google Scholar.

Reasoning

I’m interested in general and efficient goal-directed search algorithms to solve hard problems, possibly through discovery of new insights.

Large Language Models

My research focused on post-training LLMs at Amazon: SFT, RL and evaluations for coding assistants. I also worked on RAG for repo-level coding and reducing hallucinations.

Representation Learning

I had a fantastic time working on unsupervised representation learning algorithms at Adobe Research. I’ve also worked on interpretability for these models.

* Equal contribution.