Thomas Wolf

Thomas Wolf didn’t begin his career in artificial intelligence. He trained as a quantum physicist, worked for years as a European patent attorney, and built open-source tools in his spare time. It was this after-hours tinkering – writing machine learning models and sharing them publicly – that eventually sparked Hugging Face, the company that would become a cornerstone of the global open-source AI ecosystem.

Born and raised in France, Thomas studied theoretical physics at École Polytechnique before earning a PhD from Sorbonne, where he focused on quantum and statistical field theories. After a decade in law, he was drawn back to coding – not as a job, but as a creative outlet. In interviews, he’s described programming during that period as a form of intellectual play – an escape that kept his technical curiosity alive alongside his legal career.

In 2016, he teamed up with Clément Delangue and Julien Chaumond to co-found Hugging Face. What started as a chatbot app soon evolved after the team released some of their internal NLP tools as open source. The response from the developer community was immediate – and overwhelming. That pivot laid the foundation for Hugging Face’s now-flagship Transformers library, which became the default interface for working with models like BERT, GPT and T5.

Today, Hugging Face serves more than one million users and powers hundreds of thousands of AI projects across industry and academia. As Chief Science Officer, Thomas leads the company’s scientific direction, with a focus on reproducibility, transparency and open collaboration. He was a driving force behind the BigScience initiative, a year-long global research effort that culminated in the release of BLOOM, one of the largest multilingual open-access language models ever built.

He is also co-author of Natural Language Processing with Transformers (O’Reilly), a bestselling handbook for AI practitioners, and a frequent speaker at forums on responsible AI. His work consistently calls for more equitable access to advanced tools, more community-driven development, and deeper scrutiny of the societal implications of machine learning. In one recent interview, he highlighted a growing concern: that current AI systems are better at compliance than creativity – useful, but unlikely to generate the next scientific breakthrough without a shift in how models are developed and trained.

Outside of Hugging Face, Thomas is known for side projects that blend technology and imagination. One of his personal favourites – a “magic sandbox” built for his child using augmented reality – quietly went viral and sparked interest among edtech creators. He now splits his time between Paris and Amsterdam, continuing to advocate for a future where AI isn’t just powerful, but open, collaborative and human-centred.