Learn how AI actually works. Then build it.

A free, complete course on AI engineering. Start with what a large language model really is, end with shipping production AI agents. 110+ structured lessons, runnable code, an in-browser playground. No setup. No paywall.

0+
lessons
0+
tools covered
Free
forever
01 / Curriculum

From first principles
to production systems.

Start with what AI actually is. Then how it learns. Then how it talks. Every lesson uses an everyday example first, then shows you the math and the runnable code. By the end you can read a research paper without flinching.

next up
How computers turn words into numbers
embeddings · vector space · king − man + woman = queen
learnllm.dev/learn/beginnerlesson 03 of 25
Foundations

Deep Learning & Neural Networks

A stack of filters. Each layer transforms the input a little more abstractly than the last.

"golden retriever"
layer 3 — objects
layer 2 — shapes
layer 1 — edges
pixels · audio · text
02 / Agents

Build agents that
actually do things.

Modern AI doesn't just chat. It reads files, hits APIs, writes code, and ships work on its own. Build agents like that with Google ADK, OpenAI Agents SDK, CrewAI, AutoGen, and Claude — wired into GitHub, Slack, and your own systems through MCP.

tools available
github.list_open_issues
github.search_closed
github.post_comment
slack.notify
filesystem.read
triage-agent.pyrunning
> "Triage today's GitHub issues"
Thinking: I'll read open issues, classify each, then draft replies.
Tool: github.list_open_issues()
Found 7 open. 3 bugs, 2 feature requests, 2 questions.
Tool: github.search_closed_issues(similar)
3 matching bugs found. Building context...
3 drafts ready. Awaiting approval.
03 / Playground

Every lesson runs
live in the browser.

Skip the API keys, the install errors, the version mismatches. Open an in-browser playground from any lesson, edit the prompt, change the temperature, and watch the response stream back. Llama 3 70B is wired up today; Claude, GPT, and Gemini are next.

recent runs
· extract citations from paper
· fewshot classify tickets
· summarize meeting notes
playground · claude-opus-4-7temp 0.7max 2048
prompt
Write a haiku about attention
mechanisms in transformers.
response
Queries seek their keys
Softmax attention weights sing
Context flows like rain
47 tokens·0.8s·$0.0003

Three levels. One path.

110+ lessons that take you from "what is a token" to shipping production AI agents. Each level ends with a hands-on project. Self-paced. Always free.

Beginner

No code required
AI Fundamentals & How LLMs Actually Work
Prompt Engineering (CoT, Few-Shot, Patterns)
No-Code Tools (ChatGPT, Claude, Gemini)
Real-World AI (NotebookLM, Bolt.new, Cursor, Agentic AI Basics)
Outcome
Use AI like a pro, write prompts that actually work, and finally understand what's happening behind the chat box.
Start the Beginner path

Intermediate

Build real AI apps
Transformer Architecture & Modern LLMs (GPT, BERT, LLaMA)
LangChain, LlamaIndex & AI Frameworks
RAG Systems & Vector Databases (Pinecone, ChromaDB)
Agentic Frameworks (Google ADK, OpenAI SDK, CrewAI)
Modern Stack (MCP Protocol, Vercel AI SDK, AI Dev Tools)
Outcome
Build AI apps that use tools, answer questions over your own data, and run multi-step workflows end to end.
Start the Intermediate path

Advanced

Production systems
Advanced Transformers, LLaMA, Mistral & MoE
Fine-Tuning (LoRA, QLoRA, RLHF, DPO)
AI Agents (ReAct, Tool Use, Multi-Agent Systems)
Production RAG, Deployment (vLLM) & Optimization
Production Agentic (AutoGen, Orchestration, MCP Servers, Evaluation)
Outcome
Fine-tune your own models, ship multi-agent systems, and run AI in production at scale.
Start the Advanced path

Frequently asked questions.

Still have questions? We're here to help!

Contact Support

Start learning today.

Sign up free in 30 seconds. Your progress saves across every device. No credit card. No course fees. Ever.