Short Biography for Talks

Stephan Mandt is an Associate Professor of Computer Science and Statistics at the University of California, Irvine, Director of the UCI Center for Machine Learning, and co-director of UCI's AI in Science Institute. His research contributes to the foundations and applications of generative AI, with a focus on diffusion models, uncertainty quantification, neural compression, resource-efficient inference, and AI-driven scientific discovery. He is a Chan Zuckerberg Investigator and AI Resident, NSF CAREER awardee, and Kavli Fellow.

Long Biography

Stephan Mandt is an Associate Professor of Computer Science and Statistics at the University of California, Irvine. His group develops methods for generative AI, probabilistic machine learning, neural compression, uncertainty quantification, and AI for science.

The Mandt Lab studies the mathematical and algorithmic foundations of modern generative models, with a focus on diffusion models, efficient inference, and scientific data modeling. The group also applies these methods to Earth system science, chemistry, biology, physics, and other domains where uncertainty and data efficiency matter.

Stephan is Director of the UCI Center for Machine Learning, co-director of UCI's AI in Science Institute, a Chan Zuckerberg Investigator and AI Resident, an NSF CAREER awardee, a Kavli Fellow of the U.S. National Academy of Sciences, and a Fellow of the German National Academic Foundation. He regularly serves as a Senior Area Chair for NeurIPS, ICML, and ICLR, and recently served as Program Chair for AISTATS 2024 and General Chair for AISTATS 2025.

Before joining UCI, Stephan directed the machine learning group at Disney Research in Pittsburgh and Los Angeles. He previously held postdoctoral positions at Princeton University and Columbia University, and earned his Ph.D. in Theoretical Physics from the University of Cologne as a recipient of the German National Merit Scholarship.

His research has been supported by NSF, DARPA, IARPA, DOE, Disney, Intel, Qualcomm, and the Chan Zuckerberg Initiative. He has served as an Action Editor for the Journal of Machine Learning Research and Transactions on Machine Learning Research, organized tutorials at NeurIPS, AAAI, and UAI, and held visiting roles including Google Brain and ETH Zurich.

Research Areas

Foundation Models and Test-Time Control

We develop principled algorithms for generative modeling, often connecting diffusion models with stochastic optimal control. This perspective has led to methods that improve test-time behavior for deblurring, super-resolution, stylization, and related conditional generation tasks.

Uncertainty and Inference

We design inference methods that help AI systems quantify what they do and do not know. This includes Bayesian and variational approaches for deep generative models, with applications in conditional generation, scientific forecasting, and high-stakes prediction problems.

Neural Compression

We study compression as a generative modeling problem. Our work includes diffusion-based codecs, lossless compression models, information theory, and benchmarks for scientific datasets such as astronomical and climate data.

AI for Science

We collaborate with domain experts to build generative and probabilistic models for Earth system science, chemistry, biology, and physics. These collaborations both solve applied scientific problems and expose new foundational questions in machine learning.

Prospective Students

PhD applicants: Due to the volume of inquiries, I usually cannot respond individually to questions about PhD openings or application status. If you indicate your interest in working with me in the UCI application, I will review your application carefully. Please use the UCI online application page.

UCI undergraduate students: When reaching out about research opportunities, please include your resume and UCI transcript, and briefly describe which research directions interest you most. Students should have excelled in CS 178, ideally with additional coursework in AI or machine learning.

Selected News

Tutorials