AI-Infra-from-Zero-to-Hero

Why we included this project

This is a curated map of the research that powers production ML infrastructure. It collects systems papers from venues like OSDI, NSDI, SIGCOMM, and MLSys, links each one to its code when available, and organizes them into tracks for data processing, training, inference, and general infrastructure. The LLM-focused sections cover training and serving separately, while domain-specific entries reach into video systems, edge AI, graph neural networks, and federated learning. Because this is a reading list rather than software you run, its value is as a syllabus: read the white papers, work through the linked courses and video tutorials, and let it guide you from systems fundamentals to how current industry practice actually works. For an engineer or researcher building their own understanding of AI infrastructure, this is a solid place to start.