Danica J. Sutherland

Danica J. Sutherland is an Assistant Professor in the Department of Computer Science at the University of British Columbia (UBC), holding a Canada CIFAR AI Chair and affiliation with Amii (Alberta Machine Intelligence Institute). She leads research in machine learning theory, optimal transport, kernel methods, and fairness in AI. Previously, she was affiliated with the Toyota Technological Institute at Chicago (TTIC), the Gatsby Computational Neuroscience Unit (UCL), and completed her PhD at Carnegie Mellon University under Jeff Schneider.

Education: PhD in Machine Learning (Carnegie Mellon University, 2016), B.A. in Computer Science (Swarthmore College, 2011). Her work spans theoretical foundations of deep learning, active learning, generative models, and applications of optimal transport. She co-developed the POT Python Optimal Transport library and advocates for trans rights through initiatives like Queer in AI.

Research focuses on statistical learning theory, kernel-based methods for hypothesis testing, and scalable graph neural networks. Recent work includes studies on LLM finetuning dynamics, privacy-preserving generative models, and bias in language models. She teaches advanced courses on statistical learning theory and machine learning, emphasizing theoretical rigor and practical applications.

Awards include an Outstanding Paper at ICLR 2025 and a Best Paper at ACM FAccT 2023. Her lab collaborates widely, with projects funded by CIFAR, NSERC, and industry partnerships. She advises a team of PhD and master’s students, prioritizing mentorship for queer and marginalized scholars.

Labs/Teams: PI of the UBC Machine Learning Theory (ML-TN) group, co-lead of the Kantorovich Initiative (optimal transport research), and core member of the Canadian AI Institute (CAIDA). She actively contributes to open-source tools like POT and promotes inclusive AI through community initiatives.