Launch an assessment
Buy a per-candidate assessment and name the person you want to evaluate. No seats, no contracts—just one assessment at a time.
How ScreenStack works
One focused task. A complete picture.
ScreenStack gives every candidate a real codebase, a realistic 45-minute problem, and full access to AI. Then it turns what happens into an automated scorecard—what they built, and how they built it.
For hiring managers
From launch to scorecard, in three steps.
The candidate works the task
A private, ready-to-go environment with a real codebase, a focused ticket, and AI access—the same shape of work they'd do on the job.
Receive the scorecard
Every submission is scored automatically: functional results, AI direction, verification behavior, and the quality of the final change.
The scorecard
You'll see how they work with AI—not just what they produce.
Every assessment is scored automatically across four dimensions: the observable behaviors that predict how a candidate performs on a modern, AI-assisted team.
Functional outcome
The tests define “done.” If the work is correct under real conditions—including the tricky edge cases—it shows up here; if it isn't, so does the gap.
AI direction & steering
How precisely the candidate sets context, constrains the problem, and course-corrects the assistant when it drifts or hallucinates.
Verification behavior
Do they run the checks, read the diff, and verify AI-generated work before accepting it—or merge on vibes?
Quality of the final change
A small, focused, reviewable change a real team could ship—or an unchecked pile that needs cleanup first.
The candidate experience
An assessment your best candidates won't ghost on.
Strong engineers have options, and they notice how they're treated. Every ScreenStack task is designed to respect their time and mirror how modern software is actually made.
- A real ticket on a real codebase. A focused problem in a working service—never a synthetic puzzle.
- AI welcome, not policed. Candidates use their AI coding tool of choice; the assessment measures how well they direct it.
- 45 minutes, timeboxed. No weekend sacrifice, no boilerplate, no environment wrangling.
- Zero setup. Everything is provisioned and running before the clock starts.
- Transparent criteria. Candidates know exactly what they're evaluated on and how their work will be scored.
Ready when you are