Welcome to Embeddings & Semantic Search, the fifth module of the Generative AI & LLM Engineering course. So far you’ve worked with models that read and write text. This module is about a different superpower: turning text into numbers that capture meaning. An embedding is a list of numbers — a vector — that represents a piece of text so that similar meanings end up close together in space. That single idea powers search, recommendations, clustering, and the retrieval systems you’ll build in the next two modules.

You’ll start with the intuition: why “How do I reset my password?” and “I forgot my login” should land near each other even though they share almost no words. Then you’ll generate real embeddings in Python — locally and for free — with sentence-transformers, and see the actual 384-dimensional vectors. You’ll learn to measure similarity with cosine similarity and distance, and finally build a semantic search engine that answers questions over a real FAQ corpus by meaning, not keyword matching.

Every example runs for real on your own machine — no API key and no cost, because the embedding model runs locally. (We’ll also point out hosted options like Voyage AI for when you want managed, larger models.) By the end you’ll have a working search engine and the foundation for vector databases and retrieval-augmented generation, the two modules that come next.

Start with Lesson 1, where you’ll build the intuition for what an embedding actually is.