A vector database stores data as lists of numbers called embeddings, which lets it search by meaning instead of matching exact keywords. It's the piece behind a lot of recommendation systems, RAG apps, and AI agents.
We just published a comprehensive course on the freeCodeCamp.org YouTube channel that will teach you how vector databases work, starting with the math and ending with a working semantic search project. This course is from codeEdx, whose instructor has worked as a software engineer for 15 years and as a lead AI engineer for the last four. No prior knowledge is required.
In this course, you will learn:
- Vectors & Similarity Math
Learn what a vector is and how cosine similarity and Euclidean distance measure how close two pieces of data are. - Tokens & Tokenization
See why AI models can't read raw text and how a tokenizer splits a word like "indivisible" into more than one token. - Embeddings & Embedding Models
Understand how text becomes vectors that capture meaning, including the classic king minus man plus woman equals queen example. - Generating Embeddings with Gemini
Follow a live demo using Google's Gemini embedding model, then visualize the results in 3D with principal component analysis (PCA). - Inside a Vector Database
Explore the storage, indexing, and query processing layers, and how IVF and HNSW indexes avoid slow brute force search. - Semantic Search with ChromaDB
Create a Chroma client and collection, store your embeddings, and run queries that return results based on meaning.
Watch the full course on the freeCodeCamp.org YouTube channel (1.5-hour watch).
