Learn vector databases free: a 1.5-hour hands-on course that ends in a working semantic-search project

If you have worked with ordinary databases, you are used to exact matches: a query returns rows where a field equals a value. Vector databases add a different kind of search, one based on meaning rather than exact wording.

What an embedding actually is

An embedding is a list of numbers that represents a piece of text. Texts with similar meaning tend to get similar numbers, so the database can compare them. When you search, your question is turned into numbers in the same way, and the database returns the items whose numbers are closest to it. That is similarity search, and it is why a search for "how to reset a password" can surface a help article titled "recovering account access" even though the words do not match exactly.

What the course builds

The free course linked below is a hands-on introduction to this idea. It walks through storing text as embeddings, querying by similarity, and putting those pieces together into a working semantic-search project: you give it a small collection of documents, ask it a question in plain language, and it returns the most relevant documents. Following along in a code editor makes the steps easier to try and change.

Start with the free course here: https://shortun.com/yQGKj

If you need a free editor to follow along, VS Code is a solid choice: https://shortun.com/VNmwbnf

Who this is for

This is a good fit if you already write some code and want a practical first project with embeddings, or if you have heard about vector search and want to see how it works end to end rather than only reading definitions. It is less useful as a first programming course: you will get more from it if you are comfortable opening a project, running code, and reading error messages.

A sensible way to work through it

1. Watch a section, then pause and try the step yourself before moving on.
2. Start with a tiny document set so you can check each result by hand.
3. Try a few questions that are phrased differently from your documents, and see which ones the search handles well and which ones it misses.
4. Once the basic project works, change one thing at a time: add documents, try a different query, or adjust how many results you return.

That last habit matters. Semantic search is not magic, and it will return the closest matches even when none of them is a great answer. Checking results by hand on a small set is how you learn where it helps and where exact search is still the better tool.

If you finish the project, you will have something concrete to show: a small search tool that finds documents by meaning, plus a clearer sense of when a vector database is the right choice for a real application.