XNews

Stanford CS 329Z course teaches engineering AI agents from model to system

Stanford CS 329Z course teaches engineering AI agents from model to system

Stanford's Fall 2026 CS 329Z, taught by Diyi Yang, Michael Ryan, and John Yang, covers the full pipeline from simple LLM flows to autonomous agents. The course is structured around three challenges: decomposition, data, and evaluation. Modules span building primitives, frameworks and design patterns, optimization, data and evaluation, and safety and frontiers. Assignments include building an agent from scratch (Part A with litellm, Part B with DSPy) and designing a full evaluation suite for a given agent.

Source posts