finds.dev

microsoft/AI-For-Beginners

12 Weeks, 24 Lessons, AI for All!

View on GitHub

★ 70,389Jupyter NotebookMITupdated Oct 2026

// the take

A free 24-lesson, 12-week curriculum from Microsoft covering classic and foundational AI: symbolic reasoning/expert systems, neural nets, CV, NLP through transformers/BERT, genetic algorithms, RL, and AI ethics. Aimed at people starting from zero who want a structured, notebook-driven path rather than scattered tutorials.

Unusually broad for a beginner resource — it includes GOFAI/knowledge representation and genetic algorithms alongside deep learning, which most 'AI for beginners' content skips entirely. Every major lesson has parallel PyTorch and TensorFlow notebooks, so you can compare the same concept in both frameworks instead of committing to one. It's backed by real infrastructure: CodeQL scanning, OpenSSF scorecard, 50+ actively machine-translated languages, and a quiz app with per-lesson question banks. Labs force you to apply each concept to a new problem rather than just re-running the lesson's notebook.

The NLP track tops out around 2021-era material — word2vec, RNNs, and a light treatment of BERT/transformers — with one thin lesson on LLMs and prompting; there's nothing on fine-tuning, RAG, or agentic workflows, so you'll need Microsoft's separate generative-ai and agents repos to get current. It's a static notebook collection with no CI running the notebooks themselves, so PyTorch/TensorFlow API drift can silently break older lessons with no signal to users. Several lab links point to an old `microsoft.github.io` GitHub Pages mirror rather than the repo itself, which is a sign of inconsistent upkeep across lessons. As a curriculum rather than a library, there's no versioning story — you're pinned to whatever dependency versions existed when each notebook was last touched, with environment.yml files that may not reflect what's actually importable today.

We read the README, the directory tree and a key source file in October 2026. Nobody pays to be listed. How we judge →

// at a glance

Stars
70,389
Forks
13,686
Language
Jupyter Notebook
License
MIT
Last commit
23 Oct 2026

// nearby

If AI-For-Beginners isn't quite it.

All repos →