Two graduate-grade courses at William & Mary. One on building with generative models. One on engineering the systems they live inside. Both taught the same way — in the open, in front of a small room, with the projector on.
A capstone-driven studio. Fifteen scholars. Twelve weeks. Each cohort ships a real product built with — and partly by — large language models, and learns to defend every decision the model didn't make for them.
A smaller, research-shaped seminar. Papers on the table, repositories on the screen, the field still moving under our feet. We read what was published last week, then we argue about whether it actually works.
Antonio Mastropaolo is an Assistant Professor of Computer Science at William & Mary, where he leads the AURA Lab. He teaches Generative AI for Software Development (CSCI 455 / 555) — a double-listed undergraduate and graduate seminar on building with code-aware language models. His research investigates how LLMs reason about source code: evaluation, hallucination, automated program repair, and the empirical methods that tell us when a model is actually helping.
Production
William & Mary
Department of Computer Science
Williamsburg, VA
Programme
CSCI 455 / 555 — Generative AI for SD
AI for Software Engineering — Grad Seminar
Office hours by appointment