About me
Hi!
I am an AI Research Scientist at Meta Superintelligence Labs - FAIR (Fundamental AI Research) in Paris. I completed my PhD at FAIR and École polytechnique, advised by Chuan Guo and Alain Durmus, and previously by Alexandre Sablayrolles and Pierre Stock.
My research focuses on privacy preserving machine learning, data protection, and watermarking, from Differential Privacy to media provenance. I am a core contributor to Meta Seal, Meta’s official provenance offering.
Prior to my PhD I studied at École polytechnique, majoring in computer science and mathematics. I also hold a Master’s degree from École Normale Supérieure Paris-Saclay where I studied Mathematics for vision and learning (MVA). I previously interned at Polytechnique Montreal Université de Montreal, on Reinforcement Learning for Constraint Programing, and at BNP-Paribas where I developed tools for fraud detection. I also spent 6 months as a civil servant in Hanoï (Vietnam) working with the director of USTH, a scientific university.
Featured Work
News
| Feb 10, 2026 | I joined Meta FAIR as a full-time AI Research Scientist! |
|---|---|
| Feb 9, 2026 | 🎓 I defended my PhD thesis “Data Tracing in Deep Learning for Privacy and Watermarking” at École polytechnique on February 9, 2026! LinkedIn post. |
| Dec 20, 2025 | New preprint: “How Good is Post-Hoc Watermarking With Language Model Rephrasing?” is now on arXiv! |
| Jun 1, 2025 | Our work “Rethinking the Role of Verbatim Memorization in LLM Privacy” was accepted at NeurIPS 2025! |
| Feb 24, 2025 | Our work “Detecting Benchmark Contamination through Watermarking” was accepted at the ICLR 2025 Workshop! |
| Jan 22, 2025 | Our work “Watermark Anything with Localized Messages” was accepted at ICLR 2025! Paper, code and models are available. |
| Nov 12, 2024 | Our work “Watermark Anything with Localized Messages” is on arxiv! Paper, model and code are available. |
| Nov 12, 2024 | Our work “Watermarking Makes Language Models” got a spotlight at Neurips! Paper and code are available. |
| Jun 1, 2024 | Our work “Differentially Private Representation Learning via Image Captioning” is accepted at ICML 2024 in Vienna! Code and models are released! |
| Mar 4, 2024 | Our work “Differentially Private Representation Learning via Image Captioning” is on arxiv! |