CMU Launches New Course 10-749: AI for Scientific Computing with Open Lecture Notes

CMU professor Andrej Risteski is teaching a new course this semester, 10-749 "AI for Scientific Computing," focusing on the use of learned surrogates in scientific computing, covering areas such as PDE solving, forecasting, and climate prediction. Lecture notes will be gradually uploaded to the course page for public download.

Confirmed

• Risteski introduced the course himself on Twitter on September 4; its theme centers on applying learned surrogates to solving and forecasting problems in scientific computing.
• Each module follows a three-part structure: (1) mathematical foundations of classical, non-AI methods; (2) ML methodology fundamentals and "industry tricks"; (3) a survey of benchmarks in the area and the state of deployment in industry (based on public information).
• The notes assume only calculus, linear algebra, probability, and basic machine learning background—no more specialized knowledge is required—and will be uploaded incrementally to the course page for open download.
• The course is deliberately not named "AI for Science"—Risteski feels the term means different things to different people; he also explicitly states he will not cover agentic workflows or automated AI scientists.

Why it matters

• Risteski describes his motivation for writing the notes as "public service": to give people with a classical ML/AI background who want to enter these fields the mathematical grounding they need—essentially the introductory material he wishes he had when teaching himself. Given that each of these topics is an independent research area where mastering any one is nearly equivalent to a PhD, this distilled, self-study-friendly set of notes meaningfully lowers the entry barrier for newcomers to AI for Science.

sourcerefs note: all 6 posts in this cluster are first-hand publications from Risteski, the course's lead instructor.

2026-09-04 ~ 2026-09-04 · 6 related posts

• [source] CMU launches 10-749 AI for Scientific Computing: PINNs, neural operators, climate diffusion models — risteski_a · 2026-09-04
• Why the course avoids the name 'AI for science' — and skips agentic AI scientists — risteski_a · 2026-09-04
• [source] Course modules split into classical math foundations, ML tricks, and deployment surveys — risteski_a · 2026-09-04
• Professor's public-service goal: lecture notes he wishes he had while self-teaching — risteski_a · 2026-09-04
• CMU prof distills PhD-level AI for Science topics into accessible lecture notes — risteski_a · 2026-09-04
• [source] AI for Scientific Computing lecture notes to be posted, assuming only ML basics — risteski_a · 2026-09-04