Education
Andrej Karpathy (former Tesla AI director, OpenAI co-founder) walks you through building neural networks from the ground up: backpropagation by hand, micrograd, makemore, and finally a full GPT implementation. The series is unmatched for developing genuine intuition about how LLMs actually work. All code is in Python with no abstraction libraries — by the end you will have built a character-level language model and a transformer from scratch. Free on YouTube.