Learn Python
Discover courses to master Python. Browse AI-powered courses, track your progress, and earn verifiable certificates.
Available Courses
Building AI Agents: A Hands-On Guide for Developers
A hands-on Python course for developers who want to build reliable AI agents. Starting from a single model call, you'll build a tool-calling agent loop, then add well-designed tools, retrieval with citations, memory, an evaluation harness, prompt-injection guardrails and cost controls, and ship a production-ready customer-support agent. Every code sample is a real, tested file.
RAG and Vector Search: Build Search That Answers
A hands-on Python course for developers who want AI answers grounded in their own documents. You'll build a company handbook assistant step by step: chunk documents, embed them with a free local model, search by meaning and keyword, merge results with reciprocal rank fusion, rewrite follow-up questions, re-rank, and write answers with checked citations. Then you'll measure everything with recall and MRR, and ship it with permissions, a versioned index and logs. Every code sample is a real, tested file, and every number quoted comes from running it.
Building with MCP: Connect AI to Your Tools and Data
A hands-on Python course for developers who want AI apps to use their tools and data through the Model Context Protocol. You'll build an MCP server for a team task board with the official Python SDK, design tools models use well, add resources and prompts, write clients and tests, connect the tools to a model with approval for destructive actions, serve it remotely over HTTP with OAuth scopes, and defend against tool poisoning and prompt injection. Every code sample is a real, tested file.
Evaluating LLM Applications: From Vibes to Evidence
A hands-on Python course for developers and AI engineers who want to know, not guess, whether their LLM features work. Using a real support-ticket triage system, you'll build a labelled test set, write code graders and model-judge rubrics, calibrate the judge against human labels, use confidence intervals and paired comparisons to tell real improvements from noise, gate changes in CI, and monitor quality in production. Every code sample is a real, tested file.