Barret Zoph lands at Google as VP of research after brief OpenAI stint
Barret Zoph, who co-founded Thinking Machines and was later fired, has joined Google as vice president of research. He previously returned to OpenAI for five months before leaving in June. At Google, he will apply his reinforcement learning and post-training expertise to the Gemini models.
Zoph’s career path now includes three major AI employers in under two years. After leaving OpenAI in late 2024, he co-founded Thinking Machines with Mira Murati, only to depart abruptly in January alongside Luke Metz. Reports later confirmed he was fired, and his subsequent five-month return to OpenAI ended in June. At Google, he rejoins a company where he previously worked, now tasked with applying reinforcement learning and post-training methods to the Gemini model line. His hiring underscores the intense competition for AI research talent, even as executive turnover remains high across the industry.
The move also highlights a pattern of rapid leadership churn, particularly at OpenAI, which has seen departures of its COO and a top data center executive in recent months. Zoph’s experience—spanning startup co-founding, a brief return, and now a corporate research role—reflects the fluid, sometimes turbulent nature of senior roles in the AI sector, where strategic pivots and personnel shifts are common.
This appointment could signal a consolidation of expertise at major tech firms, potentially accelerating Gemini’s development in reinforcement learning and post-training. For the broader AI industry, frequent executive moves may create instability in project direction, affecting product timelines and investor confidence. Users of AI services could see indirect effects, such as shifts in model capabilities or corporate priorities. However, the impact remains uncertain, as individual researchers rarely dictate overall outcomes in large organizations.