Learn AI engineering. Build real applications. Completely free.
Free courses, hands-on exercises, and projects for building AI applications.
Learning paths
View all pathsBuild your first AI app
Call a model API, stream the reply, and get back output your code can rely on.
3 courses · 85 min of lessons · exercises additional
Prerequisite: No prior LLM experience
What it covers
- What you are actually calling
- Calling a model from your own code
- Prompting that holds up
Build a RAG system
Chunk and embed your documents, retrieve the passages that answer a question, and generate from them.
- Embeddings and similarity search
- Build a RAG pipeline
2 courses · 55 min of lessons · exercises additional
Prerequisite: Comfortable calling an API
Build an agent
Give a model tools, write the loop that calls them, and expose your own over MCP.
- Writing an agent loop
- MCP end to end
2 courses · 60 min of lessons · exercises additional
Prerequisite: Build your first AI app first
Ship reliable AI
Evaluate before you ship, then trace failures, watch cost and latency, and constrain what reaches users.
- Knowing whether it works
- Running it in production
Practice a skill
All exercisesPractice one skill at a time, with hints and a worked solution.
Validate structured model output
Take a tool call that satisfies its schema and still says something false, and catch it in code.
Beginner45 minChunk a document three ways
Split the same document by size, by structure and by section, then compare what retrieval returns.
Intermediate90 minWrite a tool definition a model can use
Start from a vague tool nobody calls correctly and rewrite it until selection is reliable.
Intermediate60 min
Build a project
ProjectsCombine several skills into an application that works end to end.
Document question answering
A working app that ingests documents, retrieves the relevant passages and answers with citations you can check.
Assumes you have read: Chunking, Storing in a Vector DB, Retrieval Process, Generation
View projectWhat you practice
- Ingesting and chunking real documents
- Retrieval scoped to one user
- Answering with citations that resolve
- Refusing when the corpus has no answer
Latest resources
205 resources- Chunking strategies that survive productionStructural, recursive and semantic splitting compared, with the sizes to start from for short lookups and for prose. Chunk size sets a ceiling on what retrieval can return, and several common mistakes lower it further.Chunking
- Building effective agentsAnthropic on the difference between workflows and agents, five composable patterns, and why most teams should start with the simplest thing that works.anthropic.comExternal
- Stop estimating tokensCharacter and word counts disagree with real token counts, which is a problem because every limit you hit is counted in tokens. How byte pair encoding splits text, and how to budget a request so a response is not cut off mid-sentence.Tokens