Buildingsuperintelligence.

I started the research at OpenAI that taught ChatGPT to think before it answers. It opened a new path to scaling AI. Today I’m a founding researcher at Meta’s TBD Lab.

My work is finding the next path to scaling intelligence.

Selected work

Earlier work spans machine learning foundations, multi-agent reinforcement learning, and learning from limited supervision. It includes work at OpenAI on competitive self-play and on continuous adaptation, which won an ICLR 2018 Best Paper.

My PhD at UMass Amherst, with Andrew McCallum, was on meta-learning without labeled data, and included research internships at OpenAI, Google Research, Microsoft Research, and Facebook.