Barret Zoph, former post-training lead at OpenAI and co-founder of Thinking Machines Lab, has joined Google as Vice President of Research. Google confirmed the appointment, stating that Zoph will direct reinforcement learning (RL) and post-training initiatives across the Gemini model family.
Background and Industry Trajectory
Zoph previously spent several years at Google Brain, where he co-authored foundational papers on Neural Architecture Search (NAS) and scaling mixture-of-experts (MoE) architectures, including ST-MoE and Z-loss regularization techniques.
In 2022, Zoph joined OpenAI, where he led post-training research teams responsible for aligning frontier models, including GPT-4 and reasoning systems. He departed OpenAI in October 2024 to co-found Thinking Machines Lab alongside former OpenAI Chief Technology Officer Mira Murati and researcher Luke Metz. In January 2026, Zoph returned briefly to OpenAI to lead enterprise AI initiatives before stepping down in June.
Focus on Gemini Post-Training and Reinforcement Learning
At Google, Zoph returns to research leadership, reporting within Google's AI division. A company spokesperson noted that his work will focus directly on post-training alignment, reasoning architectures, and reinforcement learning strategies for upcoming Gemini iterations.
Post-training methodologies, including reinforcement learning from verifiable rewards (RLVR) and process reward modeling, have become central to frontier model capabilities as pre-training scaling faces data constraints and higher compute costs. Zoph's return to Google aligns with the lab's acceleration of reasoning-focused models, following recent deployments of Gemini 3.7 Flash and Gemini Omni updates.
