Vector Search Internals
The retrieval foundation under every RAG and many agents — done right.
A deep dive into the layer most engineers treat as a black box. Why dense retrieval works at all, embedding-model selection on real data, ANN algorithms (HNSW + IVF + PQ), choosing a vector database, chunking strategies that survive production, hybrid retrieval + reranking patterns, and the eval set that catches silent decay. Capstone walks a multilingual help-center search end to end.
7h
Duration
8
Lessons
1.2k
Learners
Course map
Lessons unlock as you complete the previous one. Your progress is saved on this device.
Lesson 1
Why vector search is the right primitive
9m35 XPLesson 2
Embedding models — what to actually use
11m40 XPLesson 3
ANN algorithms — HNSW, IVF, PQ
11m45 XPLesson 4
Choosing a vector database
10m40 XPLesson 5
Chunking strategies that survive production
10m40 XPLesson 6
Hybrid retrieval and reranking — the production pattern
10m45 XPLesson 7
Evaluation — precision, recall, faithfulness
9m35 XPLesson 8
Capstone — a production-grade retrieval pipeline end to end
12m50 XP
Take next
Courses that pair well after — or alongside — Vector Search Internals.