two tracks · one idea

Learn it by seeing it, then building it.

For software engineers moving into AI/ML — every concept is something you can click, drag and break.

No hand-waving, no walls of text. Pick where you're headed. 100% free, works offline, no trackers.

live demo · self-attention Open the full lesson →

Q·Kᵀ → softmax → ·V, running live on five real 2-D toy vectors — the same three matrices every transformer runs at every layer.

▾ choose your track
🧠
Engineer path

AI/ML Engineer Roadmap

Zero → senior AI/ML engineer across 10 phases: Python, the math (as pictures), classic ML, deep learning, transformers, LLMs & RAG, MLOps, and agentic systems — design, train, deploy, operate.

244 explainers12 boss challenges11 phases + 4 modules34 quiz setsThe Model Foundry 🤖
Start here → or see the roadmap hub
🏛️
Certification prep

Claude Certified Architect

Prep for Anthropic's CCAR-F exam: the agent loop, tool & MCP design, Claude Code workflows, prompt engineering and context management — all 5 domains as playable explainers.

26 explainers5 exam domains50 quiz QsThe Agent Hangar 🚀
Enter the hangar →
6 of 244 explainers
Decision trees explainer panel Decision trees Backpropagation explainer panel Backpropagation Inside a CNN explainer panel Inside a CNN RoPE and KV cache explainer panel RoPE & KV cache RAG pipeline explainer panel RAG pipeline Attention Is All You Need illustrated paper panel Attention Is All You Need

New here? The AI/ML roadmap assumes no prior ML. The Architect track assumes you already build with LLMs.

Your place is saved in this browser — The Path picks up where you left off.

AI/ML roadmap Architect → The Hangar 📖 Illustrated Books 📱 Get the app GitHub ↗

The whole core is free forever. Optional extras — uploaded books and an AI study tutor — help it pay for itself without ads or trackers.