// series · 4 parts
RAG in Python: Zero to Production
Build a RAG system in Python from scratch — no framework. Four functions that chunk, embed, store, and retrieve over your own docs, then answer with an LLM.
SK
Sukhveer Kaur
Updated Jul 03, 2026
~27 min
Total read
Intermediate
Level
4 parts published
// the path
01
Build a RAG System in Python From Scratch (Part 1)
Build a RAG system in Python from scratch — no framework. Four functions that chunk, embed, store, and retrieve over your own docs, then answer with an LLM.
8 min
✓ Published02
RAG Chunking & Retrieval Quality: Fix Bad Answers (Part 2)
Fix RAG chunking to stop wrong answers — use fixed-size windows, overlap, and a 10-line retrieval eval. The model was never the problem; the chunks were.
6 min
✓ Published03
Semantic Chunking & Re-Ranking for Better RAG (Part 3)
Semantic chunking and cross-encoder re-ranking break the retrieval ceiling fixed-size chunks hit — with a hit-rate eval showing which to reach for first.
6 min
✓ Published04
Vector Database for RAG: When to Ditch the List (Part 4)LATEST
A vector database for RAG earns its place when persistence, filters, or scale demand it — move the Part 1 list store to Chroma and add hybrid search.
7 min
✓ Published