AI analysis grounded in the code graph — computed facts, not vibes · 2026-10-05T03:03:24Z
This is a curriculum, not a library: a structured, 503-lesson course covering AI engineering from linear algebra through to autonomous agent systems, organised into 20 sequential phases. It's built for learners moving from math foundations (autodiff, tensor operations) up through deep learning, transformers, and production agent engineering, with lesson code spanning Python, TypeScript, Rust and Julia. Each lesson pairs a written explanation with runnable code and a "reusable artifact" (a prompt, skill, agent, or MCP server), reflected in the graph as 9,838 symbols across phase directories like phases/19-capstone-projects (3,936 symbols) and phases/14-agent-engineering (745 symbols).
The README cites concrete usage figures — 150,639 readers and 241,669 page views in the last 30 days — which, alongside 39,206 stars accrued since a March 2026 creation date, suggests active promotion and word-of-mouth pickup rather than organic long-tail growth. No tagged releases were fetched, and the last 30 days of commits show only a chore(site): rebuild data.js and a README update — so recent star growth (191 in a single day) isn't explained by shipped feature work; it's more likely driven by external sharing (e.g. the author's existing following from the agentmemory project) or the "84% of students use AI but only 18% feel prepared" framing resonating with a broad audience.
What changed recently, how it's actually built (from the code graph), and whether you should care. Free account — no card, no spam.