The best free resources,
hand-picked.

mlabonne / llm-course

The de-facto LLM learning roadmap: Fundamentals, LLM Scientist, and LLM Engineer tracks with free Colab notebooks.

Best for: Phases 2–4 — your main map

microsoft / generative-ai-for-beginners

Microsoft's 21-lesson GenAI curriculum — prompting, RAG, agents, fine-tuning — with Python and TypeScript samples.

Best for: Phases 1–3 — structured basics

microsoft / ai-agents-for-beginners

18 lessons taking you from agent basics through planning, multi-agent patterns, security, and deployment.

Best for: Phase 4 — agents, end to end

Shubhamsaboo / awesome-llm-apps

100+ working open-source AI agents and RAG apps you can run, read, and rebuild. Apache-2.0 licensed.

Best for: Phases 3–4 — learn by dissecting

anthropics / claude-cookbooks

Anthropic's official notebooks: RAG, tool use, vision, sub-agents, and evaluation patterns with Claude.

Best for: Phases 2–5 — production patterns

openai / openai-cookbook

OpenAI's example gallery for the API — structured outputs, embeddings, evals — also browsable at cookbook.openai.com.

Best for: Phases 2–5 — production patterns

DataTalksClub / llm-zoomcamp

A free 10-week course on real-life LLM applications — 7 modules plus a capstone project, run as a cohort every year.

Best for: Phases 3–5 — structure + deadline

rasbt / LLMs-from-scratch

Implement a ChatGPT-like model in PyTorch, step by step — the companion code to Raschka's book.

Best for: After Phase 5 — understand the internals

chiphuyen / aie-book

Resources for Chip Huyen's AI Engineering (2025) — the book that defined this role. Chapter notes, prompts, case studies.

Best for: All phases — the reference book

microsoft / AI-For-Beginners

12-week classic-AI curriculum (neural nets, CV, NLP) with quizzes and labs — if you want broader foundations first.

Best for: Phase 1 — optional deeper base

Karpathy — Deep Dive into LLMs like ChatGPT

The single best explanation of the full LLM pipeline: pretraining, fine-tuning, RLHF, hallucinations, and how to use models well.

Best for: Phase 1 — watch this first

Karpathy — Intro to Large Language Models

The famous "busy person's intro" — what LLMs are, scaling, tool use, and security, in one lecture.

Best for: Phase 1 — the short version

Karpathy — Neural Networks: Zero to Hero

Build neural nets and a GPT from scratch, in code, from backprop up. Pairs with the nn-zero-to-hero repo.

Best for: Phase 1–2 — if you like learning by building

3Blue1Brown — Neural Networks

The most beautiful visual explanations of neural nets, gradient descent, and how transformers/attention work.

Best for: Phase 1 — intuition before code

StatQuest with Josh Starmer

~300 short videos that demystify the stats and ML concepts everyone else assumes you know. BAM.

Best for: Phase 1 — fill any gap fast

freeCodeCamp.org

Free full-length courses — search the channel for RAG, LangChain, or "LLM course" for multi-hour hands-on builds.

Best for: Phases 2–4 — long-form practice

AI Engineer (conference)

Talks from the AI Engineer Summit/World's Fair — how real teams build agents, evals, and RAG in production.

Best for: Phases 4–5 — how the industry does it

roadmap.sh — AI Engineer

A clickable skill tree for the role — track your progress topic by topic. From the team behind the famous developer roadmaps.

Best for: All phases — progress tracking

Hugging Face Learn

12 free community courses — the LLM Course and the Agents Course (free certificate) are the standouts.

Best for: Phases 2 & 4

DeepLearning.AI courses

Andrew Ng's 1–2 hour short courses with practitioners from OpenAI, Anthropic, and LangChain — RAG, agents, evals, and more.

Best for: Phases 2–5 — one topic per lunch break

fast.ai — Practical Deep Learning

The legendary top-down course: train real models in lesson one, understand the theory later. Free book included.

Best for: Phase 1 — optional ML depth

Anthropic Academy

Free certificate courses on AI fluency, the Claude API, MCP, and Claude Code — plus the hands-on anthropics/courses notebooks.

Best for: Phases 2, 4, 5 — API, tools, evals

Lil'Log — Lilian Weng

The most-cited technical explainers on agents, hallucination, and RLHF — written by an OpenAI research leader.

Chip Huyen

Essays on real-world AI engineering — her "Agents" and "Building a Generative AI Platform" posts are required reading.

Simon Willison's Weblog

Near-daily, hype-free tracking of every model release, price change, and security issue that matters.

The Illustrated Transformer

Jay Alammar's visual walkthrough of the transformer — used in courses at Stanford, MIT, Harvard, and CMU.

Latent Space

The newsletter/podcast for AI engineers — swyx coined the role. Interviews with the people building the tools you use.

Ahead of AI — Sebastian Raschka

Research digested into readable engineering takeaways — what actually changed this month and why it matters.