Teaches how to build and evaluate AI agents, from hand-written RAG and tool loops to memory, optimization, and safety. Two applied assignments and a quarter-long project connect design decisions to measured reliability.