Summary
Pierre Fernandez is a research scientist specializing in AI security and safety, with seven years of experience focused on watermarking and moderation at Meta FAIR in Paris. He led development of invisible watermarking used in high-traffic demos and production systems—supporting generation of ~1M images/day and Instagram moderation—bridging research and deployable engineering. A first author in top venues (ICCV, ICASSP, ICML, ICLR), he combines strong academic training (PhD work at Meta/INRIA, Masters from MVA, École Polytechnique and École des Ponts) with practical systems experience. He has hands-on expertise fine-tuning and evaluating large vision models (e.g., DINOv2) and shipping features that directly impact product safety at scale. Colleagues describe him as a researcher who thinks end-to-end: from novel algorithms to production integration and real-world moderation metrics.
7 years of coding experience
Mathematics and Computer Science Engineering, Mathematics and Computer Science Engineering at École des Ponts ParisTech
X2017, X2017 at Ecole Polytechnique
MVA (Masters), MVA (Masters) at Université Paris-Saclay
PhD, PhD at University of Rennes 1
Baccalauréat et Classes préparatoires, Baccalauréat et Classes préparatoires at Collège Stanislas Paris
Spanish, French, English