What makes it different
Evaluated on real work, not puzzle performance.
Your actual screen
Sessions run via native screen share. Work in your IDE, terminal, and browser — the environment you use to ship real software.
14 rubric dimensions
Scored independently across architecture, execution quality, testing, communication, system design, code reasoning, and more.
Rubrics updated weekly
The evaluation engine recalibrates against production-grade signals every 7 days. AI-assisted workflows are evaluated, not penalized.
Full transcript, always
Your session is transcribed and scored with per-category breakdowns. Review it yourself, share it with employers, or use it to improve.
Expert-calibrated baselines
Automated scoring is anchored against human expert judgment from UC Berkeley researchers and senior engineers at frontier AI labs.
AI-era aware
Using Copilot, Cursor, or Claude in your workflow? So does every strong engineer. We evaluate how you leverage AI — not whether you avoid it.
Process
How a Reeval evaluation works.
Share your screen
Start a live session. You'll share your screen and work in your own environment — your editor, your terminal, your browser. No setup required on our end.
Work on a real engineering problem
You'll receive a scope-appropriate project problem. You're scored on how you break it down, the decisions you make, how you communicate your reasoning, and the quality of your execution.
Receive your Elo and full breakdown
Your Elo score updates on the leaderboard. You get a complete transcript with per-category scoring across all 14 rubric dimensions. Share it as a verified skill credential with any hiring team.
Community
The leaderboard is a community, not a competition.
Benchmark your growth
Every evaluation updates your Elo. Re-evaluate as you improve and track your trajectory across rubric categories over time.
Get discovered
Hiring teams browse the public leaderboard filtered by rubric category and Elo tier. A strong Reeval score is a verifiable credential.
Track the bar
As rubrics recalibrate weekly, the leaderboard reflects shifts in industry expectations. You'll know when the bar moves — and what it moved to.