FlagOS OpenCourse

Open educational resources for the FlagOS open-source AI system software stack — lectures, hands-on labs, tutorials, and teaching guides covering the full stack from AI computing hardware fundamentals to compilers, high-performance operators, and distributed training.

🎓 The online learning center will be available at https://edu.flagos.io

Repository Structure

This repository is organized around one main-line course, from which everything else is derived:

course/ is the mainline — the complete, actively developed and iterated curriculum. All ongoing course development happens here.
editions/ holds releases derived from the mainline, tailored to specific audiences and programs.
best-practices/ holds hands-on component tutorials that complement the course.

Directory	Role	Description
`course/`	Mainline	高校合作支持中心（中文）：48 课时模块化课程课件、实验材料包、课程大纲示例、教师培训指南、在线实验室与高校合作流程。持续开发迭代的课程主体。
`editions/tencent-edu/`	Release	Tencent education platform edition (English): syllabus, general education course, five textbook modules, and hands-on labs
`editions/china-africa-faculty/phase-1/`	Release	China–Africa AI Compute Faculty Development Program, Phase 1: syllabus, general education course, textbook modules in English and French, and runnable Ascend labs
`best-practices/`	Tutorials	Hands-on tutorials for the four core FlagOS components (FlagGems, FlagTree, FlagScale, FlagCX)

Releases under editions/ are organized by program, and each program by phase (phase-1/, phase-2/, …), so future cohorts are added alongside existing ones without disturbing them.

University Course (中文课程体系)

A comprehensive 48-lecture modular course for universities, covering the full AI system software stack:

Module	Topic
1	AI 系统软件基础与异构计算 / AI Systems Software Foundations & Heterogeneous Computing
2	高性能 AI 算子与算子工程 / High-Performance AI Operators & Operator Engineering
3	AI 编译器原理与优化 / AI Compiler Principles & Optimization
4	分布式并行训练与通信 / Distributed Parallel Training & Communication
5	性能评测与下一代内核生成 / Performance Benchmarking & Next-Gen Kernel Generation

Chinese and English slides, homework, and lecture video indexes are provided under course/02-课件资源/. Universities can access online labs with GPU/NPU compute support — no local hardware required. See 08-高校合作流程/ for how to participate.

Component Best Practices

In-depth technical guides for each core FlagOS component (principles, case studies, and hands-on practice), under best-practices/:

Component	Description
FlagGems	High-performance general-purpose AI operator library (Triton-based)
FlagTree	Unified AI compiler for multi-chip backends
FlagScale	Large-scale distributed training and inference framework
FlagCX	Unified cross-chip communication library

Labs

Hands-on lab materials (NPU & Triton basics, performance tuning, and LLM deployment) are available under the labs/ directory of each program, with environment check scripts and reference implementations.

Contributing & Contact

Issues and suggestions: GitHub Issues
FlagOS community: https://flagos.io · Discord

License

Course materials in this repository are licensed under CC BY-NC 4.0 (Creative Commons Attribution-NonCommercial 4.0 International).