![\"AI](\"assets/banner.svg\")
\n\n \n
\n Read in your language:\n Español ·\n Français ·\n Português ·\n Deutsch ·\n Italiano ·\n 简体中文 ·\n 日本語 ·\n 한국어 ·\n हिन्दी ·\n العربية ·\n Русский ·\n Türkçe\n
Translated landing pages, committed to the repo. English is canonical; lesson pages are machine-translated on the translations branch. See docs/i18n.md.\n
\n ![\"MIT](\"https://img.shields.io/badge/license-MIT-1a1a1a?style=flat-square&labelColor=fafaf5\"){kind=icon}
![\"523](\"https://img.shields.io/badge/lessons-523-3553ff?style=flat-square&labelColor=fafaf5\"){kind=icon}
![\"20](\"https://img.shields.io/badge/phases-20-3553ff?style=flat-square&labelColor=fafaf5\"){kind=icon}
![\"GitHub](\"https://img.shields.io/github/stars/rohitg00/ai-engineering-from-scratch?style=flat-square&labelColor=fafaf5&color=3553ff\")
![\"Website\"](\"https://img.shields.io/badge/web-aiengineeringfromscratch.com-3553ff?style=flat-square&labelColor=fafaf5\"){kind=icon}
![\"GitHub](\"https://img.shields.io/github/stars/rohitg00/agentmemory?style=flat-square&labelColor=fafaf5&color=3553ff\")
which naturally works with any agents or chat assistants.\n\n```text\n░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒\n```\n\n> **84% of students already use AI tools. Only 18% feel prepared to use them\n> professionally.** This curriculum closes that gap.\n>\n> 523 lessons. 20 phases. ~342 hours. Python, TypeScript, Rust, Julia. Every lesson ships\n> a reusable artifact: a prompt, a skill, an agent, an MCP server. Free, open source, MIT.\n>\n> You don't just learn AI. You build it. End-to-end. By hand.\n\n\n114,584 readers · 181,995 page views in the last 30 days · as of 2026-08-29
\n\n\n## Start here: choose what you want to build\n\nYou do not need to scan 523 lessons before beginning. Pick one goal. Each link\nopens the same curriculum on GitHub or the website, and both versions use the\nsame lesson code.\n\n| Your goal | Learn on GitHub | Learn on the website |\n|---|---|---|\n| I am new and want the complete foundation | [Phase 0: Setup and Tooling](phases/00-setup-and-tooling/) | [Dev Environment](https://aiengineeringfromscratch.com/lesson?path=phases/00-setup-and-tooling/01-dev-environment) |\n| I know Python and want math plus ML foundations | [Phase 1: Math Foundations](phases/01-math-foundations/) | [Linear Algebra Intuition](https://aiengineeringfromscratch.com/lesson?path=phases/01-math-foundations/01-linear-algebra-intuition) |\n| I want to build production LLM applications | [Phase 11: LLM Engineering](phases/11-llm-engineering/) | [Prompt Engineering](https://aiengineeringfromscratch.com/lesson?path=phases/11-llm-engineering/01-prompt-engineering) |\n| I want to build agents | [Phase 14: Agent Engineering](phases/14-agent-engineering/) | [The Agent Loop](https://aiengineeringfromscratch.com/lesson?path=phases/14-agent-engineering/01-the-agent-loop) |\n| I want to use coding agents on real repositories | [Agent-Assisted Engineering path](learning-paths/using-coding-agents.json) | [Agent-Assisted Engineering](https://aiengineeringfromscratch.com/lesson?path=phases/14-agent-engineering/31-agent-workbench-why-models-fail&learningPath=using-coding-agents) |\n| I want to shape the right build before implementation | [Product Judgment and Delivery path](learning-paths/shaping-the-build.json) | [Product Judgment and Delivery](https://aiengineeringfromscratch.com/lesson?path=phases/14-agent-engineering/47-outcomes-before-output&learningPath=shaping-the-build) |\n| I want to build with Model Context Protocol (MCP) | [Model Context Protocol (MCP) route](phases/13-tools-and-protocols/README.md#model-context-protocol-mcp-path) | [Model Context Protocol (MCP) path](https://aiengineeringfromscratch.com/lesson?path=phases/13-tools-and-protocols/06-mcp-fundamentals&learningPath=model-context-protocol) |\n| I want to write and ship Agent Skills | [Focused Agent Skills route](phases/13-tools-and-protocols/README.md#agent-skills-fast-path) | [Agent Skills path](https://aiengineeringfromscratch.com/lesson?path=phases/13-tools-and-protocols/22-skills-and-agent-sdks&learningPath=agent-skills) |\n| I want to prepare for a Claude certification | [Certification onboarding](certifications/claude/GETTING_STARTED.md) | [Certification Academy](https://aiengineeringfromscratch.com/certifications.html) |\n\nNot sure where you fit? Use the [`start-learning` placement tutor](skills/start-learning/SKILL.md)\nor the [website prerequisites guide](https://aiengineeringfromscratch.com/prereqs.html).\n\nCompare four core domains and six career routes in the [AI Engineering Learning Paths](https://aiengineeringfromscratch.com/learning-paths.html).\n\n### Use every lesson the same way\n\n1. **Read** `docs/en.md` and explain the core idea in your own words.\n2. **Type and build** the important code instead of treating the code block as decoration.\n3. **Run** the lesson command from the repository root, the directory containing `README.md` and `phases/`.\n4. **Keep evidence**: the command, working directory, exit code, meaningful output, and the artifact you changed or produced.\n5. **Continue** only when you can explain the output and make one small change without guessing.\n\nCommands in lesson pages are paths from the repository root unless the lesson\nexplicitly says to change directories. If a lesson offers several languages,\nrun the implementation for the language you are learning.\n\n### Clone it and produce your first evidence\n\n```bash\ngit clone https://github.com/rohitg00/ai-engineering-from-scratch.git\ncd ai-engineering-from-scratch\npython3 phases/00-setup-and-tooling/01-dev-environment/code/verify.py --route beginner\npython3 phases/01-math-foundations/01-linear-algebra-intuition/code/vectors.py\n```\n\nThe preflight separates requirements needed now from tools needed later. Every\nrequired failure includes the detected reason and a corrective command. The\nsecond command is a dependency-free lesson and ends by showing that a matrix\ntimes a vector is the operation inside a neural network layer. Save that\nterminal output as your first evidence.\n\n## Add the AI tutor in 30 seconds\n\nIf Node.js, `npx`, and a skill-capable coding agent are already installed,\nyour coding agent can become your tutor in two commands. A repository clone is\nnot needed to install or read the tutor. Runnable focused-path labs need\n`python3`. Agent Skills host labs also need a selected host and a writable\nuser or project skill scope.\n\nCheck the local requirements first:\n\n```bash\nnode --version\nnpx --version\npython3 --version\n```\n\nThen install the curriculum skills and choose the host and scope you intend to\nuse when the installer asks:\n\n```bash\nnpx skills add rohitg00/ai-engineering-from-scratch\n```\n\nInvocation syntax belongs to the host, not to the portable `SKILL.md` format:\n\n| Host | Start the course | Start Model Context Protocol (MCP) | Start Agent Skills | Run a phase quiz |\n|---|---|---|---|---|\n| Codex | `start-learning`, or choose it from `/skills` | `learn-mcp`, or choose it from `/skills` | `learn-agent-skills`, or choose it from `/skills` | `check-understanding 13`, or choose it from `/skills` |\n| Claude Code | `/start-learning` | `/learn-mcp` | `/learn-agent-skills` | `/check-understanding 13` |\n| Other compatible hosts | `Use start-learning to begin the course.` | `Use learn-mcp to start the Model Context Protocol (MCP) path.` | `Use learn-agent-skills to start the Agent Skills Engineering path.` | `Use check-understanding to quiz me on Phase 13.` |\n\nA ten-question placement quiz maps what you already know to a starting phase and\nsaves a personalized study plan to `LEARNING.md`. From there, the `learn` skill\nteaches one lesson per session: concept, math, code, quiz. It streams lessons\nstraight from this repo, and the `course-guide` skill jumps you to the exact\nlesson that covers anything you are stuck on. In Codex, invoke these skills with\n`learn` and `course-guide`; in Claude Code, use `/learn` and `/course-guide`;\nin other compatible hosts, ask to use the skill by name.\n\nOnly want Model Context Protocol (MCP)? Use the MCP invocation for your host. It creates\n`MCP-LEARNING.md` and follows one 17-lesson route through stateless\nrequests, transports, bidirectional work, security, reliability, registry\ngovernance, and conformance evidence. The exact order and checkpoints live in\nthe [Model Context Protocol (MCP) manifest](learning-paths/model-context-protocol.json).\n\nOnly want Agent Skills? Use the Agent Skills invocation for your host. It\ncreates `AGENT-SKILLS-LEARNING.md` and follows one coherent five-lesson route:\ncontract, discovery, invocation, sandbox boundaries, then release evals and\nreal-host portability. Start on the web with the\n[Agent Skills path](https://aiengineeringfromscratch.com/lesson?path=phases/13-tools-and-protocols/22-skills-and-agent-sdks&learningPath=agent-skills).\n\nThe installer lists the hosts it can configure and asks where to install. If\nyou do not have Node.js, `npx`, `python3`, a supported host, or a writable\nscope yet, use the website or read `docs/en.md` manually. That path teaches the\nconcepts, but real-host discovery, invocation, script, and uninstall evidence\nremains pending until the preflight is available. Read the lessons at\n[aiengineeringfromscratch.com](https://aiengineeringfromscratch.com).\n\n## How this works\n\nMost AI material teaches in scattered pieces. A paper here, a fine-tuning post there, a\nflashy agent demo somewhere else. The pieces rarely line up. You ship a chatbot but can't\nexplain its loss curve. You hook a function to an agent but can't say what attention does\ninside the model that's calling it.\n\nThis curriculum is the spine. 20 phases, 523 lessons, four languages: Python, TypeScript,\nRust, Julia. Linear algebra at one end, autonomous swarms at the other. Every algorithm\ngets built from raw math first. Backprop. Tokenizer. Attention. Agent loop. By the time\nPyTorch shows up, you already know what it's doing under the hood.\n\nEach lesson runs the same loop: read the problem, derive the math, write the code, run\nthe test, keep the artifact. No five-minute videos, no copy-paste deploys, no hand-holding.\nFree, open source, and built to run on your own laptop.\n\n```text\n░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒\n```\n\n## The shape of the curriculum\n\nTwenty phases stack on top of each other. Math is the floor. Agents and production are the roof.\nSkip ahead if you already know the lower layers, but don't skip and then wonder why something at\nthe top is breaking.\n\n```mermaid\n%%{init: {'theme':'base','themeVariables':{'primaryColor':'#fafaf5','primaryTextColor':'#1a1a1a','primaryBorderColor':'#3553ff','lineColor':'#3553ff','fontFamily':'JetBrains Mono','fontSize':'12px'}}}%%\nflowchart TB\n P0[\"Phase 0 — Setup & Tooling\"] --> P1[\"Phase 1 — Math Foundations\"]\n P1 --> P2[\"Phase 2 — ML Fundamentals\"]\n P2 --> P3[\"Phase 3 — Deep Learning Core\"]\n P3 --> P4[\"Phase 4 — Vision\"]\n P3 --> P5[\"Phase 5 — NLP\"]\n P3 --> P6[\"Phase 6 — Speech & Audio\"]\n P3 --> P9[\"Phase 9 — RL\"]\n P5 --> P7[\"Phase 7 — Transformers\"]\n P7 --> P8[\"Phase 8 — GenAI\"]\n P7 --> P10[\"Phase 10 — LLMs from Scratch\"]\n P10 --> P11[\"Phase 11 — LLM Engineering\"]\n P10 --> P12[\"Phase 12 — Multimodal\"]\n P11 --> P13[\"Phase 13 — Tools & Protocols\"]\n P13 --> P14[\"Phase 14 — Agent Engineering\"]\n P14 --> P15[\"Phase 15 — Autonomous Systems\"]\n P15 --> P16[\"Phase 16 — Multi-Agent & Swarms\"]\n P14 --> P17[\"Phase 17 — Infrastructure & Production\"]\n P15 --> P18[\"Phase 18 — Ethics & Alignment\"]\n P16 --> P19[\"Phase 19 — Capstone Projects\"]\n P17 --> P19\n P18 --> P19\n```\n\n```text\n░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒░░░▒▒▒\n```\n\n## The shape of a lesson\n\nEach lesson lives in its own folder, with the same structure across the entire curriculum:\n\n```text\nphases/![\"FIG_001.A](\"site/assets/figures/001-a-prompts.svg\") FIG_001 · A PROMPTS | \n![\"FIG_001.B](\"site/assets/figures/001-b-skills.svg\") FIG_001 · B SKILLS | \n![\"FIG_001.C](\"site/assets/figures/001-c-agents.svg\") FIG_001 · C AGENTS | \n![\"FIG_001.D](\"site/assets/figures/001-d-mcp-servers.svg\") FIG_001 · D MCP SERVERS | \n
|---|---|---|---|
| Paste into any AI assistant for expert-level help on a narrow task. | \nDrop into Claude, Cursor, Codex, OpenClaw, Hermes, or any agent that reads SKILL.md. | \nDeploy as autonomous workers — you wrote the loop yourself in Phase 14. | \nPlug into any MCP-compatible client. Built end-to-end in Phase 13. | \n
| \n\n**`code/agent_loop.py`** build it\n\n```python\ndef run(query, tools):\n history = [user(query)]\n for step in range(MAX_STEPS):\n msg = llm(history)\n if msg.tool_calls:\n for call in msg.tool_calls:\n result = tools[call.name](**call.args)\n history.append(tool_result(call.id, result))\n continue\n return msg.content\n raise StepLimitExceeded\n```\n\n | \n\n\n**`outputs/skill-agent-loop.md`** ship it\n\n```markdown\n---\nname: agent-loop\ndescription: ReAct-style loop for any tool list\nphase: 14\nlesson: 01\n---\n\nImplement a minimal agent loop that...\n```\n\n**`outputs/prompt-debug-agent.md`**\n\n```markdown\nYou are an agent debugger. Given the trace\nof an agent run, identify the step where\nthe agent went wrong and explain why...\n```\n\n | \n
22 lessons The intuition behind every AI algorithm, through code.18 lessons Classical ML — still the backbone of most production AI.13 lessons Neural networks from first principles. No frameworks until you build one.28 lessons From pixels to understanding — image, video, 3D, VLMs, and world models.29 lessons Language is the interface to intelligence.17 lessons Hear, understand, speak.16 lessons The architecture that changed everything.15 lessons Create images, video, audio, 3D, and more.12 lessons The foundation of RLHF and game-playing AI.24 lessons Build, train, and understand large language models.17 lessons Put LLMs to work in production.25 lessons See, hear, read, and reason across modalities — from ViT patches to computer-use agents.31 lessons The interfaces between AI and the real world.