Alan Sun

About Me

I am a first-year MSCS student at Carnegie Mellon University.

This past summer, I was a visiting scholar at the Max-Planck Institute for Software Systems advised by Mariya Toneva. I earned my undergraduate degree in Computer Science and Mathematics at Dartmouth College with highest honors. At Dartmouth, I conducted research with Soroush Vosoughi where I worked to create a mathematical framework that enables the characterization of robustness of language models. My honors thesis, Acheiving Domain-Independent Certified Robustness via Knowledge Continuity, was awarded First Prize for Oustanding Research in Computational Sciences by the Neukom Institute as well as the John G. Kemeny Computing Prize for Innovation.

My current research focuses on interpreting the mechanisms of artificial neural networks through circuit discovery.

Research

I am interested in improving the robustness and interpretability of deep learning models by uncovering the mechanisms underlying their decision-making processes and developing frameworks to enhance their stability. My research explores how models reason and adapt across tasks, introducing concepts like knowledge continuity and circuit stability to provide a unified understanding of robustness. By bridging gaps between theory and practice, I aim to expose the opaque links between adversarial vulnerabilities, interpretability challenges like superposition, and the structural properties of neural representations, enabling safer and more reliable AI systems.

Publications