AISystem

Why we included this project

AISystem is a structured open course on the AI systems stack, and its value is that it explains the layer most framework tutorials skip: how AI actually runs on hardware. The modules cover AI chip architecture (CPU, GPU, NPU, and dedicated processors from NVIDIA, Google, and Chinese vendors), compilers from front-end optimizations to kernel back-ends, inference engines, and framework internals such as automatic differentiation and computation graphs. Each module comes with slides, written lessons, and video, so it works as a self-study path for graduate students and as a reference for engineers who want to understand the systems layer beneath the frameworks they use daily. The author has moved newer large-model content into a separate repository, which keeps this course focused on the fundamentals.