I've noticed that semantic search tutorials only use cloud vector databases and embedding APIs, so I figured I'd take a stab at writing a simpler guide. It's possible to get good accuracy and speed entirely locally.