The pipeline behind the atlas.

01Sources are watched, not scraped

The pipeline starts with a video the editor actually learned from. Its transcript is fetched; the material is broken into concepts, prerequisites, misconceptions, exercises, and quotable moments with timestamps. A source registry records every ingested video and why it qualified.

02Synthesis, not summary

Notes from multiple sources are woven into an original guide — the atlas never republishes anyone's lecture. The writing tool reads the whole curriculum first and decides whether the material becomes a new guide, extends an existing one, or splits across several.

03The pull request is the editorial desk

Nothing publishes directly. Every draft arrives as a pull request; the editor reviews, corrects, and merges. Merging is publishing — the site rebuilds and the new star appears on its shelf.

04Learning science, applied

• Andragogy. Adults learn what they can use — every guide opens with why it matters to real work and ends with an exercise you run tonight.
• Bloom's-taxonomy objectives. The objective box promises observable verbs — estimate, restructure, diagnose — never "understand"; you can check whether you left the room with them.
• Retrieval practice. The oral examination at the end of each guide makes you pull the ideas back out of your own head — reveal-style answers, no grades, pure recall.
• Scaffolding. Prerequisite links and the Broad → Practitioner → Deep shelves let each guide stand on charted ground instead of assuming it.

05Built like it teaches

The site itself is a static Astro build: design tokens in one documented file, a custom titling typeface generated from stroke skeletons, and a WebGL star chart whose stars are the guides themselves. The craft is the credential.