OpenAI 大模型算法岗系统训练工程

This is an open-source book series for interview preparation and capability building for large model algorithm roles.

It is not a collection of scattered notes but a complete training system. It covers machine learning fundamentals, Transformer, LLM training, alignment, inference deployment, inference frameworks, AI Infra, evaluation and safety, paper close reading, project practice, system design, the tool protocol ecosystem, and mock interviews. Step by step, it helps readers build the theory, engineering, research, and communication skills that large model algorithm roles require.

The purpose of this book series is not to list knowledge points but to help readers build a transferable knowledge map of large models. For important concepts, it tries to explain the historical background, the motivation for the idea, the problems with earlier methods, the characteristics of current methods, their strengths and weaknesses, their applicable boundaries, the directions later work has improved on, and how they relate to adjacent concepts. The project is beginner-friendly by default but targets readers at every stage: beginners can follow the context and…