rag-from-scratch

Why we included this project

This tutorial builds a retrieval pipeline in plain JavaScript using only local LLMs, and it explains each step instead of hiding it behind a library call. Embedding, chunking, vector storage, retrieval, re-ranking, and query rewriting are all covered line by line, so you can read exactly how context gets retrieved and merged into a prompt. The examples follow a progressive path, which makes it a good place to build a mental model before committing to a heavier framework. Developers who have been burned by black-box tools will find it useful for debugging and tuning a real RAG setup later. Teams just starting with retrieval-augmented generation get a solid grounding without any cloud dependency.