Generative AI for Beginners (Version 3)
by Microsoft
A free 21-lesson, code-first on-ramp from prompts to RAG, function calling, agents and fine-tuning, in Python and TypeScript.
Overview
Generative AI for Beginners is an open-source course maintained by Microsoft Cloud Advocates on GitHub, now in its third version, with more than 120,000 stars, 63,000 forks, 2,400+ commits and active maintenance through September 2026. It is organised as a setup lesson plus 21 numbered lessons, each tagged either Learn (conceptual) or Build (hands-on code). The Learn track covers an introduction to generative AI and LLMs, comparing different models, responsible AI, prompt-engineering fundamentals and advanced prompts, UX design for AI applications, securing generative AI applications, the generative AI application lifecycle, fine-tuning LLMs, small language models, and building with Mistral and Meta models. The Build track has you write text-generation and chat applications, search apps with vector databases, image-generation apps, low-code apps, function-calling integrations, a retrieval-augmented generation (RAG) app, open-source model work with Hugging Face, and AI agents. The agents lesson compares LangChain agents, AutoGen, TaskWeaver, JARVIS and the Microsoft Agent Framework (the successor that merges Semantic Kernel and AutoGen), including a typed-function tool-calling example. Code samples run in Python and TypeScript against Azure OpenAI, the OpenAI API, Microsoft Foundry Models or Foundry Local, most lessons include a video introduction and a Keep Learning section, and the text is translated into 50+ languages. It is breadth-first and application-level: it teaches you to consume model APIs well, not to train models.
At a Glance
- Topic
- Frameworks
- Level
- Beginner
- Format
- Course
- Cost
- Free
- Duration
- 21 lessons plus a setup lesson, self-paced (no official time estimate; most lessons have a short intro video)
- Provider
- Microsoft
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Explain how large language models work and compare different LLM families for a task
- ✓Write effective prompts, then apply advanced prompting techniques to improve output quality
- ✓Build text-generation and chat applications against Azure OpenAI or the OpenAI API
- ✓Build semantic search and a RAG application backed by a vector database
- ✓Integrate external applications with LLMs using function calling and structured tool schemas
- ✓Compare agent frameworks including LangChain, AutoGen and the Microsoft Agent Framework
- ✓Decide when fine-tuning an LLM or using a small language model is appropriate
- ✓Secure generative AI applications and manage their lifecycle responsibly from prototype to production
Highlights
- •One of the most-starred AI courses on GitHub (120k+ stars, 63k+ forks), still actively maintained in September 2026
- •Every Build lesson ships runnable code in both Python and TypeScript, not just notebooks
- •Explicit Learn vs Build split lets you skip theory and jump straight to hands-on lessons
- •Translated into 50+ languages with short video intros for most lessons
- •Honest caveat from Hacker News: it is API-consumption focused and leans on Azure/Microsoft services, so it will not teach you to train or run models yourself
Who It’s For
Best For
- ✓Software developers building their first LLM-powered application
- ✓Python or TypeScript engineers who want a structured tour of prompts, RAG, function calling and agents
- ✓Teams standardising onboarding for developers new to generative AI
- ✓Developers already on Azure or Microsoft Foundry
Prerequisites
- •Basic Python or TypeScript programming
- •A GitHub account to fork the repository
- •An API key for Azure OpenAI, OpenAI or Microsoft Foundry (Foundry Local for offline use)
FAQ
What is Generative AI for Beginners (Version 3)?
Generative AI for Beginners is Microsoft's free, open-source course for software developers who are new to building with large language models. Across 21 lessons it moves from LLM basics and prompt engineering to building chat, search, image-generation, RAG and agent applications, so you finish able to ship a working LLM-backed app in Python or TypeScript.
Is Generative AI for Beginners (Version 3) free?
Generative AI for Beginners (Version 3) is free to access.
What level is Generative AI for Beginners (Version 3) for?
Generative AI for Beginners (Version 3) is aimed at a beginner audience. Recommended background: Basic Python or TypeScript programming, A GitHub account to fork the repository, An API key for Azure OpenAI, OpenAI or Microsoft Foundry (Foundry Local for offline use).
How long does Generative AI for Beginners (Version 3) take?
Expect roughly 21 lessons plus a setup lesson, self-paced (no official time estimate; most lessons have a short intro video). Most learners work through it at their own pace.
What will I learn from Generative AI for Beginners (Version 3)?
You'll learn: Explain how large language models work and compare different LLM families for a task; Write effective prompts, then apply advanced prompting techniques to improve output quality; Build text-generation and chat applications against Azure OpenAI or the OpenAI API; Build semantic search and a RAG application backed by a vector database; Integrate external applications with LLMs using function calling and structured tool schemas; Compare agent frameworks including LangChain, AutoGen and the Microsoft Agent Framework; Decide when fine-tuning an LLM or using a small language model is appropriate; Secure generative AI applications and manage their lifecycle responsibly from prototype to production.
Topics
Sources
This page was written from 4 sources, 1 on domains other than github.com.