Build the systems
you use.
Seven projects from task queues to inference platforms. Each one teaches what documentation can't — how real systems behave under load, failure, and changing requirements.
Seven systems, three levels each.
Each project moves through progressive → ambitious → research-grade. Build the smallest correct version first, then make it production-like, then investigate the internals.
Go deeper when something pulls you.
Advanced systems projects for after the core sequence — or when a particular area becomes irresistible. None are mandatory.
Choose your depth.
After establishing the core engineering foundation, pick one or more tracks for focused exploration. These are optional branches, not prerequisites.
Twelve steps, every project.
A repeatable workflow for starting, building, and completing each project. Each step includes a concrete example using the Distributed Task Queue, plus how to use AI as a learning partner at that stage.
Eight principles to keep it honest.
These keep the roadmap ambitious without making it chaotic.