Skip the résumé. See real code from every candidate.
Start screening engineers on real code. 10 standard invitations free.
Sample report · canned data

Sit in the reviewer's chair.

Two submissions to the same assessment. One is a genuinely strong session; one passed every test and should never be hired. Click through the evidence and make the call yourself — this is exactly what lands in your inbox after every screening.

Fully interactive
Every score opens its evidence
No signup needed
PS
Priya Sharma
Backend Engineer · event pipeline assessment
AI allowed
87
FINAL · PASS MARK 70
Advance to interview
SCORES — CLICK ONE TO OPEN ITS EVIDENCE
EVIDENCE · AI AUGMENTATION
Will merge_windows handle an empty spans list?
No — spans[0] raises IndexError. Add a guard: if not spans: return []
Is sorting spans in place safe here, or should I copy first?
The caller reuses the list later in your code — copy first (sorted(spans)) to avoid the side effect.
Verified AI output before accepting
rewrote the guard in her own style — 54% re-edited
Asked about edge cases unprompted
empty input · in-place mutation — before either failed a test
Re-ran the public tests after each AI-assisted change
2 of 2 times — green both times
SESSION TIMELINE46 min of 60
typingthinking pauseAI conversation

Steady typing with short thinking pauses; two AI consultations (cyan), both followed by her own edits. Real work has a rhythm — this is it.

THE CALLEvery number here survives inspection: steady build rhythm, edge cases asked about before they bit, AI output verified and rewritten. This is what a strong session looks like on paper.

All names and data on this page are fictional. In the real product, every flag and score row opens the keystroke replay itself.

Your turn

Run this on your own candidates.

The first 10 screenings are free — same report, your pipeline, no card.