Sebastian Stich

Sebastian Stich is a tenured faculty member at the CISPA Helmholtz Center for Information Security, where he leads research in Trustworthy Information Processing. He has been a tenure-track faculty since 2021 and was promoted to tenured professor in 2025. He is also a member of the European Lab for Learning and Intelligent Systems (ELLIS).

Education:

• PhD in Computer Science, ETH Zurich (2010–2014)
• MSc and BSc in Mathematics, ETH Zurich (2005–2010)
• Research Scientist, EPFL (2016–2021)
• Research at CORE/ICTEAM, UCLouvain (2014–2016)

His research centers on optimization for machine learning, with a focus on federated, decentralized, and distributed learning. He investigates methods for communication efficiency, adaptive stochastic optimization, privacy-preserving training, and generalization theory. His work bridges theoretical guarantees with practical scalability.

His recent publications (2023–2025) consistently address gradient compression, error feedback, local updates, and decentralized consensus, demonstrating a strong trend toward making distributed learning more efficient, robust, and scalable—especially under heterogeneous data and limited bandwidth.

Scientific Awards:

• ERC Consolidator Grant 2024 (CollectiveMinds)
• Google Research Scholar Award (2023)
• Meta Privacy-Enhancing Technologies Research Award (2022)

Sebastian Stich actively advises PhD students and postdocs, including Anton Rodomanov, Xiaowen Jiang, and Yuan Gao. He has secured competitive grants such as the ERC CollectiveMinds project, supporting collaborative research on scalable federated learning. He teaches advanced courses at Saarland University and serves as an area chair for NeurIPS, ICML, and ICLR.

He leads a research group at CISPA focused on trustworthy and efficient machine learning systems, contributing to both foundational theory and real-world applications in privacy and security.