ML Roadmap
Zero to a deployed system, one falsifiable checkpoint at a time.
Forty nodes across eight phases. Each one gives you a goal, two resources, a coding session, and a checklist you can genuinely fail. Nodes unlock when their prerequisites are passed — locked ones stay visible, so you can always see the shape of the road ahead.
00Tooling floor5 nodes
01Math on demand5 nodes
02Supervised learning7 nodes
03Unsupervised and feature work5 nodes
04Deep learning6 nodes
05Applied LLM5 nodes
06Ship it6 nodes
07Capstone1 node
All forty nodes are written. Nothing here has been through a pilot yet, so expect rough edges — and if a checklist item is wrong, that is a bug worth reporting.