cs249r_book

The Harvard CS249r repository provides an integrated, multi-volume curriculum designed to teach the discipline of AI engineering. Moving beyond model training, the project focuses on building efficient, reliable, and scalable intelligent systems. The curriculum is structured into four core volumes: Foundations (Vol I), Scaling (Vol II), Agentic Systems (Vol III), and Physical AI (Vol IV). Each volume bridges algorithmic theory with the systems infrastructure required for real-world deployment. The project includes a robust ecosystem of learning tools: - TinyTorch: A framework for building ML internals from scratch. - Labs: Interactive Marimo notebooks for exploring systems trade-offs. - MLSys·im: A simulator for calculating infrastructure bottlenecks like memory and network saturation. - Hardware Kits: Practical exercises for deploying models on edge devices like Arduino and Raspberry Pi. - StaffML: A collection of physics-grounded interview questions and practice drills. - Instructor Hub: Resources for educators, including syllabi, lecture slides, and assessment rubrics. This repository serves as both a textbook series and a practical toolkit, aiming to help learners master the intersection of machine learning and systems engineering by emphasizing quantitative reasoning, hardware constraints, and end-to-end system design.