Publications

Advancing the State-of-the-Art in Empirical Privacy Auditing
Redirection for Erasing Memory (REM): Towards a Universal Unlearning Method for Corrupted Data
Learning with User-Level Differential Privacy Under Fixed Compute Budgets
Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy, Research, and Practice
JaxPruner: A Concise Library for Sparsity Research
Examining Data Compartmentalization for AI Governance
DrJax: Scalable and Differentiable MapReduce Primitives in JAX
Towards Federated Foundation Models: Scalable Dataset Pipelines for Group-Structured Learning
Leveraging Function Space Aggregation for Federated Learning at Scale
How Federated Learning Protects Privacy