Writing.

How I built Womli's RAG pipeline — and what I'd do differently

A walkthrough of the retrieval architecture behind Womli: chunking strategy, embedding model choice, pgvector configuration, and the three decisions I'd change if I started over today.