Began working as a research scientist at Google Brain, contributing to projects like PaLM, Flan-T5, and the JAX-based large-scale training system T5X.

Joined OpenAI as a Research Scientist, focusing on reasoning and agents. Became a Foundational Contributor to o1 and led the Codex mini model-training.

Was a Foundational Contributor to the release of o1-preview, a significant project at OpenAI focused on reasoning.

Was a Foundational Contributor to the full release of o1, an OpenAI model focused on learning to reason with LLMs.

Joined Meta's Superintelligence Labs as an AI Research Scientist, moving from OpenAI with colleague Jason Wei.