Nikolay Savinov joins OpenAI to work on pretraining

Nikolay Savinov announced on X that he has joined OpenAI to work on pretraining, based in its London office. He moves over from Google DeepMind.

At DeepMind, Savinov worked on large-scale pretraining and long context: the capability that lets Gemini models read and reason over very large inputs, from hundreds of thousands up to millions of tokens at once.

His research career began further upstream. He earned his PhD at ETH Zürich, working on 3D computer vision, and went on to influential reinforcement-learning work on exploration and memory before moving into large language models.

At OpenAI, he says he will help grow the London research team, which is actively hiring.