Summary
Grégoire Delétang is a quantitative researcher with a decade of experience building large-scale deep learning systems, currently applying those skills at Squarepoint Capital after a research-engineering tenure at DeepMind. His work spans large language and sequence models, online learning, compression, and multi-agent reinforcement learning, grounded in rigorous math training from École Polytechnique, ENS Paris-Saclay (MVA), and Télécom ParisTech. He combines research-grade experimentation with production-minded engineering—shipping models and infrastructure that scale—while maintaining an academic presence (Google Scholar) and an active habit of "learning at scale" showcased on GitHub. Notably, his background includes privacy and decentralized-systems work on personal VPNs and practical experience porting deep RL to imperfect-information games, reflecting an unusually broad bridge between theory, systems, and real-world deployment.
10 years of coding experience
5 years of employment as a software developer
Master's degree, Mathématiques Vision Apprentissage (MVA), Master's degree, Mathématiques Vision Apprentissage (MVA) at ENS Paris-Saclay
Diplôme d'ingénieur, Diplôme d'ingénieur at Télécom ParisTech
École Polytechnique
Classes Préparatoires aux Grandes Écoles, MPSI - MP*, Classes Préparatoires aux Grandes Écoles, MPSI - MP* at Lycée Sainte-Geneviève
French, English, Chinese, Italian