About me

Since September 2020, I have been doing AI research at OpenAI. I started my AI research career working on pretraining data and co-wrote the first Codegen paper. My work has also included capability prediction (scaling laws for evals beyond perplexity), optimization and multi-cluster training, and long-context pretraining. I was a primary contributor to GPT-3.5, which powered the first version of ChatGPT, and GPT-4.

For the past two years, I have been leading a team focused on reasoning, memory, and test-time training.

Before joining OpenAI, I was a software engineer at Rubrik in Palo Alto, CA, working on distributed systems, cloud, and backend technologies. I earned my PhD in Theoretical CS from MIT in 2017.

contact

Feel free to reach out to me via email.