Barret Zoph’s AI executive tour has taken another sharp turn. The Thinking Machines co-founder, who was fired from the startup in January before rejoining OpenAI for a five-month stint heading enterprise sales, has now landed at Google as vice president of research. A Google spokesperson confirmed the move to the Wall Street Journal with a line that reads like a deliberate jab at the chaos elsewhere: “We look forward to Barret returning to Google and bringing his RL and post-training expertise to Gemini.”
Zoph’s trajectory over the past nine months is almost hard to follow. He spent two years at OpenAI, left in October 2024 to co-found Thinking Machines with Mira Murati, departed that startup in January along with co-founder Luke Metz to return to OpenAI, was later revealed to have been fired from Thinking Machines, then left OpenAI again in June after just five months on the enterprise sales side. That’s a lot of movement for someone now tasked with reinforcement learning and post-training work on one of the most important model families in the industry.
The churn isn’t unique to Zoph. OpenAI has been bleeding senior talent for months — its COO departed, a top data center executive recently walked, and the list goes on. For a company reportedly preparing for an IPO and sitting at the center of the AI boom, the instability is notable. Investors tend to prefer leadership continuity when billions are on the line, and every high-profile exit gives competitors a fresh opening to poach institutional knowledge.
For Google, this hire is a clear signal that the Gemini team is serious about closing whatever gap remains in post-training quality. Zoph’s background spans both research and go-to-market, and having someone who understands enterprise sales pain points while also working on RL could tighten the feedback loop between what customers actually need and how models are tuned. The real question is whether he stays long enough to matter.
