What scanning real repos reveals about how AI engineers work

Inferr never reads a line of source code — the scanner only ever sees metadata: commit history, languages, and which AI tools and frameworks actually show up in a repo. That's a narrow window, but it's a real one. Here's what it shows across every verified engineer on the platform, live.

A note on method: every figure below is computed from verified or elite profiles only — engineers whose work has cleared Inferr's corroboration bar, not just anyone who ran the scanner. Small samples are noisy; treat early numbers as directional.

How the verified pool breaks down

Every public profile lands in a tier based on the strength of its proof — from a clean scan up through independently verified production revenue.

Why this is different from a survey

Self-reported tool usage is a popularity contest — people say what sounds good. This is measured from what's actually committed to version control by engineers who passed independent verification. As the verified pool grows, this report gets more precise automatically — no new survey required.