2,212 saved links — your single biggest topic. This tag is the capture-everything reflex: every course, book, roadmap, and free-certification thread you didn’t want to lose. The queue is unfinishable by design, so this page is the opposite of the archive: the shortlist actually worth sitting down with, grouped by what kind of learning you’re in the mood for.
Related: Productivity & Learning · Career · AI (General) · Programming
Start here: how to learn (before what to learn)
- Seroter: stop following tutorials and learn by building instead and learn AI by shipping real projects, not tutorials — the antidote to this very page.
- curiosity-driven learning vs. fear-driven “upskilling” — that AI course tab you’ve had open for weeks.
- the 500-page book on the science of learning — how learning actually works, before you queue up more courses.
- “a few dozen book pages a day should become the new 10,000 daily steps”.
- The Great Mental Models series — first-principles, second-order, probabilistic thinking.
The Claude mastery track
The most-saved single thread in the whole tag — and the one you’re literally practicing right now.
- The official ladder: Claude 101 → Claude Code 101 → API fundamentals → Claude Certified Architect — Foundations.
- Cert prep: an exam preparation guide and study guides in four languages.
- Frontend Masters’ free Claude Code course (with Anthropic) and Claude Code in Action on Coursera.
- Anthropic × DeepLearning.AI: free Agentic AI Skills course and Agent Skills step-by-step from zero.
- 100+ Claude Code resources, courses and tools — the meta-list.
- Off-label uses: reading across books with Claude Code · building skills without coding (video).
AI engineering curricula
- Chip Huyen’s AI Engineering book resources and her free AI-engineering roadmap — the field’s reference syllabus.
- “a free AI university in one repo” — real-world AI systems step by step.
- AI Hero: engineering course with examples and an evals framework · DAIR’s production-ready agents course · LangChain Academy: building reliable agents.
- Microsoft’s AI Agents for Beginners — MCP, A2A, NLWeb.
- Agentic Design Patterns: a hands-on guide — free from Google and the agentic handbook: battle-tested patterns.
- Nathan Lambert’s RLHF book with the companion video course.
- the JAX scaling book — how LLMs actually run in production.
- Alisa’s Book of LLMs — an OpenAI hire’s interview-prep notes, free.
- “the AI agents roadmap nobody is teaching you”.
The free university
The classic-CS shelf — no paywalls, all of it re-readable forever.
- MIT Press opened 500+ books and 12 free MIT AI books, compiled.
- MIT: Introduction to Computational Thinking and Data Science · MIT 6.824 distributed systems — the recommended path into foundation models.
- Jeff Erickson’s “Algorithms” — the free PDF classic.
- CMU Advanced Database Systems · Helsinki’s compilers course · Algorithmica — the open web book on performance.
- Dive into Deep Learning — interactive, with code and math · Berkeley’s deep representation learning open book.
- Karpathy’s 2h13m tokenizer video, translated into a proper book chapter and course.fast.ai.
- Google’s free SRE books · the NASA Systems Engineering Handbook (297-page PDF) — your two disciplines’ canonical freebies.
Certifications
- GitHub’s new Agentic AI Developer certification (GH-600) — multi-agent orchestration, guardrails, production workflows.
- OpenAI’s first certification courses · Azure skilling challenges — 50% off a cert · CompTIA SecAI+ — free beta exam attempts.
- a quick-start guide to AI certs — read the free docs and build first, then certify.
- the free Google/IBM/HP certification list.
One-weekend wins
Small enough to finish — the cure for queue guilt.
- Pamela Fox’s 4-hour MCP workshop and “build your first MCP server in Python” from PyCon.
- an interactive git guide with quizzes.
- Qdrant Academy — interactive vector-search courses.
- Google’s 5-day self-paced agents intensive.
- free Linux Essentials course + certificate · free Linux & server management.
- codebase-to-course — turn any repo into an interactive HTML course with one instruction — very devopsmantra-adjacent.