Stanford Launches CS329Z: Engineering AI Agents from Scratch This Fall
Stanford will debut a new course, CS329Z "Engineering AI Agents," this fall, taught by Diyi Yang, Michael Ryan, and John Yang, with the course website already live. It is a landmark move to systematically bring agent engineering practice into university curricula.
Confirmed
- Per Diyi Yang's original posts, instructors are Diyi Yang, Michael Ryan, and John Yang (some reposts give divergent third-instructor names; the official site is authoritative)
- The syllabus follows a "monolithic LLM → compound AI systems → autonomous agents" arc, covering training data, RAG, agent design patterns and harnesses, and evaluation
- Two assignments: building a harness from scratch, and an evaluation challenge against frontier models
- The course launches this fall with the syllabus publicly available
Unconfirmed
- The full instructor list: reposts disagree on the third instructor's name
Why it matters
- It systematically brings the industry's hottest agent engineering practices into the classroom, covering the full stack from data to retrieval to evaluation
- Hands-on assignments target real engineering problems, offering high reference value for students and practitioners entering agent engineering
2026-09-03 ~ 2026-09-04 · 7 related posts
Primary sources
- [source] Stanford launches CS329Z, a new course on engineering AI agents from scratch — Diyi_Yang · 2026-09-03
- [source] Stanford CS329Z assignments: build an agentic harness and novel frontier-model evals — Diyi_Yang · 2026-09-03
- [source] Stanford CS329Z site is live with full syllabus on engineering agentic systems — Diyi_Yang · 2026-09-03
- Stanford's new CS329Z treats Agent Engineering as a discipline: decomposition, data, evaluation — udmrzn · 2026-09-04
3 near-duplicate retellings: Diyi_Yang · jyangballin · Diyi_Yang