Hadrien Hendrikx

Hadrien Hendrikx is a Researcher (Chargé de Recherche) at Inria Grenoble in the Thoth team. He was previously a postdoctoral researcher at EPFL (2021-2022) and completed his PhD at Inria Paris in the SIERRA and DYOGENE (now ARGO) teams, affiliated with the Computer Science Department of Ecole Normale Supérieure. He also collaborated with the MSR-INRIA joint center.

• Education:
  • BSc from Ecole Polytechnique (2016)
  • MSc in Computer Science from EPFL (2018)
  • PhD in Machine Learning under Francis Bach and Laurent Massoulié (2021)

Hendrikx’s research focuses on decentralized and distributed optimization in machine learning, emphasizing robustness, privacy, and personalization. He also explores mirror descent methods and their stochastic theory, alongside interdisciplinary applications in ecology, such as the CASCA project using computer vision for underground fauna analysis.

His recent publications span topics like Byzantine-robust gossip protocols, variance reduction in decentralized optimization, and topology-aware learning, reflecting a blend of theoretical rigor and practical ecological impact. Notable awards include an Excellence Fellowship and an Outstanding Paper Award at NeurIPS 2021.

• Supervision:
  • Current PhD students: Renaud Gaucher, Loukas Duqué, Géraud Ilinca
  • Current Master’s students: Morgan Scalabrino, Manuela Giraldo Obando
  • Former advisees: Mohamed Bacar Abdoulandhum, Daniel Morales Broton, Abdellah El Mrini, Rustem Islamov, Mathieu Even

Hendrikx teaches Numerical Optimization at Université Grenoble Alpes and part-time Generalization Properties of Machine Learning Algorithms at Orsay. He also contributes to peer review for top conferences (ICML, NeurIPS, AISTATS) and journals (Mathematical Programming, IEEE Transactions).