Summary
William Tong is a Research Assistant and PhD student at Harvard with a decade of experience at the intersection of deep learning, cognitive science, and software engineering. He develops theoretical and empirical work on neural architectures—demonstrating, for example, in-context learning in MLPs and formulating prompt-style/task-topology theory for Transformers—results published and presented at venues including ICLR and CCN Proceedings. His background spans applied research and production systems, from designing a 1.6B-token Lean proof dataset to building deployed NLP pipelines and AWS-based ingestion services. William blends rigorous mathematical training (PhD/MS in Applied Mathematics/Data Science) with hands-on engineering at Google and startups, enabling him to validate theory at scale. Notably, his work often uncovers underappreciated phenomena (e.g., equality reasoning and learning richness in MLPs) that reshape assumptions about model capabilities. Based in Cambridge, MA, he brings both experimental creativity and practical delivery to problems in reasoning and representation learning.
10 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Applied Mathematics, Doctor of Philosophy - PhD, Applied Mathematics at Harvard University
Bachelor of Arts - BA, Computer Science, Bachelor of Arts - BA, Computer Science at Columbia University in the City of New York
High School, High School at Illinois Mathematics and Science Academy