Autonomous Software Engineering

Course overview

This course studies how AI agents can support software engineering work across requirements, architecture, coding, testing, and operations.

The emphasis is on building harnesses that combine models, tools, memory, and evaluation. We will work with OpenCode, Claude Code, and GitHub Copilot, and build our own harnesses throughout the course. We will also examine where human-in-the-loop review is needed—and how people and agents can work together when judgment and accountability matter.