El Mahdi El Mhamdi

El Mahdi El Mhamdi is an Assistant Professor of Mathematics and Data Science at École Polytechnique, where he co-chairs the Master of Science program in Data Science and Machine Learning. His research focuses on the robustness of distributed learning systems and AI safety, with notable contributions to Byzantine-resilient algorithms like Bulyan and Krum. These algorithms are deployed in industrial settings such as WeBank and IBM's Federated Learning Library. He has authored four papers featured in the US NIST 2024 report on Adversarial Machine Learning as state-of-the-art defenses.

Education: Grew up in Morocco with a bachelor-level background in physics, mathematics, and philosophy. Pursued studies in France, Germany, and Switzerland through scholarships, including a PhD in Computer Science from EPFL. His thesis on Robust Distributed Learning received the 2020 Best Computer Science PhD distinction and was shortlisted for a cross-departmental award.

Research Interests: Robustness of biological networks (e.g., neural, gene-regulatory systems) and distributed machine learning systems under adversarial conditions. Advocates for ethical AI governance and has published a book on AI safety with Lê Nguyên Hoang, translated into multiple languages. His work addresses societal impacts of AI, including misinformation and data poisoning.

Awards: Includes France’s top 100 scientific inventors (2025), VentureKick award (2020), and recognition from Global Voices Online, Thomson Reuters, and Google for freedom of speech advocacy. Legal actions ongoing against Moroccan authorities due to threats and harassment.

Grants & Service: Co-authored research plans totaling $1.2M in grants. Served as a reviewer/PC member for top conferences (NeurIPS, ICML, ICLR). Supervised multiple master’s theses at EPFL, fostering talent in distributed computing and AI safety.

Labs/Teams: Active in collaborative projects like Tournesol (trustworthy human judgments) and Aksel (Byzantine-resilient SGD). Engages in public outreach through podcasts, radio interviews, and science communication initiatives.