microsoft/generative-ai-for-beginners

12 Lessons, Get Started Building with Generative AI 🔗

Summary

microsoft/generative-ai-for-beginners is a 21-lesson course (Version 3) from Microsoft Cloud Advocates that teaches engineers how to build generative AI applications, with each lesson pairing a written explanation, a short video, and Python and TypeScript code samples. The MIT-licensed curriculum moves from LLM fundamentals and prompt engineering to hands-on builds for text, chat, search, image, RAG, function calling, and AI agent applications.

What it is useful for

The repo is aimed at engineers with basic Python or TypeScript experience who want a structured path into generative AI development. The 21 lessons split into "Learn" topics (responsible AI, prompt engineering, UX for AI, security, the LLM lifecycle, fine-tuning) and "Build" topics (text generation, chat, vector search, image generation, low-code, function calling, RAG with vector databases, open-source models via Hugging Face, AI agents, plus Mistral, Meta, and SLM model families). A separate .NET edition is linked in the README for C# developers. Lessons are modular, so you can jump to RAG, function calling, or agents without working through earlier ones.

How engineers can use it

1. Start with lesson 00 (Course Setup) to configure your environment. The README lists four runtimes: Azure OpenAI Service, Microsoft Foundry Models, OpenAI API, or Foundry Local (fully offline, no cloud subscription).
2. Fork the repository to your own GitHub account (required per the README).
3. To avoid pulling all 50+ translations, use the documented sparse checkout:

   git clone --filter=blob:none --sparse https://github.com/microsoft/generative-ai-for-beginners.git
   cd generative-ai-for-beginners
   git sparse-checkout set --no-cone '/*' '!translations' '!translated_images'
   

4. Pick lessons in order, or jump straight to a "Build" lesson that matches your project—for example, lesson 08 for vector search, lesson 11 for function calling, lesson 15 for RAG, or lesson 17 for AI agents.
5. Each lesson ships parallel Python and TypeScript samples, so you can stay in the language that matches your stack.

Documented limitations to keep in mind: a few lessons (00, 19, 20, 21) are listed as "Video Coming Soon," and the README notes GitHub Models is retiring at the end of July 2026, so use Microsoft Foundry Models for those githubmodels lessons instead. Absolute beginners are directed to separate Python and TypeScript prerequisite courses before tackling this material.