I built a coding agent from scratch in Python — no frameworks, no LangChain — to understand how AI agents actually work. ReAct loops, tool use, security sandboxing, human-in-the-loop — here is every component, every design decision, explained from first principles.

Everyone uses ChatGPT. Few people understand how it actually works. I built a decoder-only Transformer language model from scratch in PyTorch — no nn.TransformerDecoder, no Hugging Face — to demystify the architecture behind GPT-2. Here is every layer, every matrix multiply, every design decision.