Matthieu Meeus
Researcher
Now Research Scientist, FAIR · Meta
@matthieu_meeusmatthieumeeus.com ↗GitHub ↗LinkedIn ↗Latest move
Imperial College London → Meta · Research Scientist, FAIR
Belgian researcher who did his computer science PhD at Imperial College London with Yves-Alexandre de Montjoye, on what large language models memorize from training data: membership inference, copyright traps and privacy audits. He first trained in mechanical engineering and worked as a data scientist.
Path
MetaResearch Scientist, FAIRNEW
Imperial College LondonPhD, Computer Science, advised by Yves-Alexandre de Montjoye2022–2026- MRMicrosoft Research CambridgeResearch Intern2024–2024
- TETeslaIntern, residential battery charge/discharge strategies
- MKMcKinsey & CompanySenior Data Scientist
- KUKU LeuvenBSc, Mechanical Engineering
UC BerkeleyMEng, Energy Technology
Harvard UniversityMSc, Computer Science
Papers
- SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)SaTML 2025 2024Reviews membership inference attacks on large language models, shows that common evaluation setups are flawed, and proposes better ones.
- The Canary's Echo: Auditing Privacy Risks of LLM-Generated Synthetic TextICML 2025 2025Builds a pipeline that audits whether synthetic text generated by a fine-tuned language model leaks the data it was trained on.
- Did the Neurons Read your Book? Document-level Membership Inference for Large Language ModelsUSENIX Security 2024 2023Trains a classifier on token probabilities to infer whether a given document, such as a book, was in a language model's training data.
SourcesLatest move: Xmatthieumeeus.comEducation and jobs are read from their own homepage or employer's page; papers checked on arXiv. From LinkedIn only the person's own public posts are read; GitHub is a link only.