CourseRAG · Module 5: Vector Stores and Indexing · part 28 of 82
Part 28 · Module 5: Vector Stores and Indexing

Topic 6: Abstraction

16 min read·21 Sept 2026

6.1 The Retriever Interface: A Vector DB Is Not a Special Case

Intuition: your RAG app should ask "give me the best passages for this question, within these limits", not "run a Qdrant HNSW query". Whether the answer comes from a vector store, a keyword index, a SQL table, or a mix is an implementation detail.

The rest of this course is yours to keep

This course is bought on its own, once, and stays readable afterwards, including the parts added to it later.