The whole territory

The Map

Five stages, no optional ones. The route runs from the first principles you actually need through to the frontier of AI engineering â€” with a research branch that forks at Stage 2 for anyone serious about going further. Every node names what Expert and Principal look like, not just Competent.

Depth at every stage â†’

AwareCompetentProductionExpert / SeniorPrincipal / Researcher

The map names what Expert and Principal look like â€” not just Competent.

Every stage

Skills that compound regardless of which stage you’re on.

Software craft
Code is read more than it is written.
Signal: You refactor before you ship, not after it breaks.
Communication
A finding that changes nothing was a waste of time.
Signal: Non-technical colleagues repeat your conclusion in their own words.
Tooling fluency
Know your instruments — they shape what you can think.
Signal: You reach for the debugger before the print statement.
Scientific thinking
State the hypothesis before running the experiment.
Signal: You design the measurement before you build the feature.
Learning to learn
Primary sources over summaries. Teaching over watching.
Signal: You read the paper, not just the tweet thread about the paper.
Ethics and responsibility
Optimising for the wrong metric is worse than not optimising.
Signal: You ask 'who is harmed by this?' before shipping anything.