Experiment in motion · The Racing Data Machine

From 200+ messy sources to 1.12 million clean pages.

Fully automated — no human gates. Chaos in, structure out.

What you're looking at

This is the real pipeline behind raceteam.wiki. Hundreds of scattered, inconsistent sources get scraped, normalized, deduplicated, enriched, validated, and deployed as more than a million static pages — all without a human in the loop. Press play and watch the data move.

Every stage runs on a schedule. The pipeline is fully autonomous: no approval gate, no operator standing by. The goal is a single clean record per team, every time, worldwide.

🔒 Flow and public script names are real. We don't list our sources or reveal our dedup heuristics — that's the moat.

agent at work work in flight