Research focuses on algorithmic foundations of modern AI—developing parallel methods to characterize neural network behavior and creating testing frameworks for generative models to enhance reliability, interpretability, and human‑centered deployment.