A Fun & Absurd Introduction to Vector Databases
summary
A hands-on build shows a Unity game front end calling a Kotlin Spring Boot backend to a local VV8 Docker instance, storing 384‑dim dense embeddings and indexing with HNSW and cosine distance; text uses a BERT embedding, images use ResNet/CLIP, audio uses Whisper. Bulk writes are slow so indexing is batched once offline. Demos include an RPG spell search, a 721‑image Pokémon matcher that often but not always finds Pikachu, icon-to-effect mapping that works, and a metal‑song match that fails.
AI-written summary · about
chapters (15)
- 0:00Intro
- 1:28Goal of this talk
- 2:54Vectors
- 5:47Machine learning
- 9:05Vector databases
- 14:53Demo
- 16:45Vector databases continued
- 19:16Implementing our own search
- 21:40Demo: HNSW
- 25:04Demo: Matrix
- 28:17Demo: D&D
- 31:49Demo: Pokemon
- 36:34Demo: D&D
- 41:55Demo: Duncan Hills
- 45:07Outro
related talks
48mA fun and absurd introduction to Vector Databases
48mA Fun & Absurd Introduction to Vector Databases
47mA fun and absurd introduction to Vector Databases
46mA fun and absurd introduction to Vector Databases
50mA fun and absurd introduction to Vector Databases
01h 23mVector Databases Demystified: Discover how they work
01h 46mVector Databases (for GenAI & beyond)
45mA gentle introduction to Vector Databases