Thibault Sellam

How to pronounce my name: Tee-Bo—it's French!

I am a Staff Research Scientist at Google Deepmind in NYC. I am interested in machine learning and LLMs. I work on the Gemini post-training team, with a strong focus on tool-use, reasoning, and factuality. I also developed evaluation methods for text and speech generation (e.g., SQuId and BLEURT), and pre-training robustness (MultiBERTs).

Before joining Google, I worked on database systems and visualization. I was a PhD student at the University of Amsterdam and CWI (the Netherlands), then a postdoc at Columbia.

I am also a music nerd—when I am not at the office, you may find me play with a band, an orchestra or enjoy one of New York's many music venues. I also like to apply ML to music, see examples here.