Pierre Stock is VP of Science at Mistral AI, Europe's leading open-weight AI company, where he serves as the first-ever employee and drives model releases, roadmap strategy, and execution.

Pierre earned a Master of Science in Machine Learning and Computer Vision from Ecole Normale Superieure Paris-Saclay, completed his PhD at Facebook AI Research exploring efficiency and redundancy in large language models, and later worked as a Research Scientist at Meta specializing in neural network compression and privacy-preserving machine learning.

Pierre is a vocal advocate for open-weight AI models as a counterweight to proprietary US and Chinese systems, arguing that open access democratizes frontier AI technology for developers and enterprises worldwide.

At Mistral AI, Pierre co-developed the landmark Mistral 7B model, co-authored the Mixtral of Experts architecture, led the release of Codestral 22B for 80+ programming languages, and spearheaded Voxtral, a state-of-the-art real-time transcription model.

Mistral AI has grown into Europe's most-funded AI startup with over $6.5B raised, and recently previewed ML4, a 1-trillion-parameter open-weight model trained on 4,000 NVIDIA Grace Blackwell GPUs in its own European data centres.