Resources that helped me transition into AI Infrastructure

This year, I made the transition into AI Infrastructure. During my transition, I spent countless hours reading books, writing code, watching videos, and exploring open-source repositories. These resources helped me tremendously, and I hope they can help others as well.

English

GPU

• NVIDIA CUDA Programming Guide — official documentation
• NVIDIA cuda-samples — GitHub repository
• Programming Massively Parallel Processors — book
• GPU MODE — YouTube channel
• cuda-oxide — GitHub repository

AI / AI Infra

• Andrej Karpathy's YouTube channel (highly recommended) — explanations of how LLM training and inference work, along with hands-on tutorials
• freeCodeCamp — AI, Machine Learning, and PyTorch courses (YouTube channel)
• vLLM — GitHub repository
• llama.cpp and the Llama source code — GitHub repository
• Research papers from leading AI companies and research labs (OpenAI, Anthropic, Meta, Google DeepMind, DeepSeek, NVIDIA, etc.)
• Kubernetes — repo and official documentation

Mandarin

AI / AI Infra

• @uncle_moon_x — YouTube channel, highly recommended — interviews with researchers and engineers at the forefront of AI

---

If you're also transitioning into AI Infrastructure or AI Systems, I hope these resources save you some time and make your learning journey a little easier.