Hanshen Xiao is a PhD candidate whose research explores the area of privacy-preserving machine learning. Specifically, he investigates information-theoretical security, robust optimization, and signal processing, from theoretical and practical perspectives. MATLAB is an indispensable tool in Hanshen’s work, supporting such tasks as large-scale spectrum analysis, testing learning algorithms, and estimating higher-order cumulants. His recent accomplishments include studies in high dimensional differentially private stochastic optimization with heavy-tailed data and task augmentation for private (collaborative) learning on transformed data. With the support of his MathWorks Fellowship, Hanshen will pursue a cutting-edge approach in private transformation- based machine learning with the goals of improving security properties and demonstrating that transformed data learning maintains utility. This work has already yielded a promising result in collaborative learning that does not require collusion assumptions unlike established multiparty computation schemes. This project, and Hanshen’s future research, have the potential to advance privacy-preserving machine learning, offering both new theoretical knowledge and practical applications.