🎮 Learn local AI by playing
Level up your AI skills
Run, ship, secure and test AI on your own hardware. Short lessons with animated diagrams, quizzes and labs, plus XP, streaks and badges to keep you going.
- lessons
- 146
- lessons
- courses
- 10
- courses
- fixes
- 15
- fixes
- free lessons
- 41
- free lessons
How the game works
1. Learn
Bite-size lessons with animated diagrams you can play with, not walls of text.
2. Practice
Quizzes after every lesson and hands-on labs you run on your own machine.
3. Level up
Earn XP, keep your daily streak alive and collect badges as you master each course.
Pick your quest
All courses →Run models on your own hardware
Local LLM Inference
Hardware sizing, quantization, runtimes, performance tuning, serving basics, edge and applications. The complete path from “what is a token” to a tuned local setup.
41 lessons19 free15 animated diagrams
From one working model to a reliable service
Serving AI in Production
What changes when real users depend on your model: service levels, capacity and cost planning, gateways, autoscaling, observability, safe rollouts and failure handling.
36 lessons8 free3 animated diagrams
Build LLM apps that are hard to abuse
AI Security and Privacy
The threat model for LLM applications, prompt injection, data leakage, tool abuse, privacy engineering and compliance basics, with a hands-on red-team lab.
12 lessons4 free3 animated diagrams
Know whether your AI actually works
LLM Evaluation and Testing
Replace vibes with evidence: golden datasets, metrics, statistics, LLM-as-judge, RAG and agent testing, regression suites in CI and production monitoring.
12 lessons4 free4 animated diagrams
Why inference is fast or slow, and how to make it faster
Inference Engineering
Bottleneck maths, attention and GPU architecture, the software stack, engines compared from the inside, and the techniques that matter: quantisation, speculative decoding, parallelism, caching and disaggregation. Ends with a benchmarking capstone.
12 lessons1 free0 animated diagrams
Tokenizer to trained transformer, with code you run
Build an LLM from Scratch
Write a BPE tokenizer, attention, a GPT block, a training loop and a sampler in plain Python first, then see how the same ideas map to PyTorch and to modern upgrades like RoPE, RMSNorm, SwiGLU and grouped-query attention.
10 lessons1 free0 animated diagrams
Measure, scale and train reasoning step by step
Build a Reasoning Model
What reasoning means and how to verify it, chain of thought, self-consistency, search, verifier loops, reinforcement learning with verifiable rewards (a GRPO-style method), distillation and honest evaluation, with runnable toy experiments.
7 lessons1 free0 animated diagrams
A tiny llama.cpp, written by you
Build an Inference Engine
Implement a complete forward pass, KV cache, sampling, quantised matrix multiplication, batching and an OpenAI-compatible server, first in readable Python and then with the C++ ideas that make it fast.
7 lessons1 free0 animated diagrams
Vision, speech, embeddings and images
Multimodal Inference
How vision-language models, embedding and reranking services, speech recognition, text-to-speech, real-time voice pipelines and diffusion models run, what they cost and how to serve each.
5 lessons1 free0 animated diagrams
Adapt a model without wrecking it
Fine-tuning and Customisation
When to fine-tune, data preparation, LoRA and QLoRA mathematics, preference tuning, distillation, merging, evaluating a fine-tune and serving adapters, with a runnable LoRA demonstration.
4 lessons1 free0 animated diagrams
Collect all 21 badges
Every lesson earns XP. Streaks, perfect quizzes, labs and finished courses unlock badges. Your progress lives in your browser, so there is nothing to sign up for.
See my progress- First Step
- Bookworm
- Scholar
- Living Legend
- Warming Up
- On Fire
- Bullseye
- Sharp Mind
- Lab Rat
- Level 5
- Local Hero
- Ship It
Structured tracks
From “what is a token” to production serving, in an order that makes sense.
Interactive tools
VRAM calculator, hardware checker, command builder, quantization explorer.
Run a model right here
WebGPU playground: chat with a real LLM on your own GPU, no install.
Fix-it wiki
Searchable solutions for CUDA OOM, slow Macs, garbled output and more.
Free forever, Pro when you're ready
Everything you need to understand and run local models is free. Go Pro for advanced tracks, labs, bigger playground models and certificates.
Compare plans- Foundations, hardware and runtime basics
- All tools, glossary and troubleshooting wiki
- Pro: advanced optimization, serving & application tracks
- Pro: hands-on labs with checklists
- Pro: larger in-browser models and certificates
FAQ
Is it really free?
Yes. All foundations, hardware, runtime and evaluation basics, every tool, the glossary and the troubleshooting wiki are free. Pro adds advanced lessons, hands-on labs, larger playground models and certificates.
Do I need a GPU?
No. Plenty of lessons apply to CPUs and Macs, and the playground runs small models in a browser. A GPU or Apple Silicon Mac just makes bigger models practical.
How current is the content?
Every lesson shows when it was last verified. Runtime flags change quickly, so we note version-sensitive details and point you to --help for your build.
Can I cancel Pro anytime?
Yes. Manage or cancel your subscription from the billing portal on your account page.