skill > resume

Hire developers by what they build.

Coding challenges, hackathons and real tasks — every candidate solves the same problem, then you interview a shortlist ranked by the work, not the paper.

Run a pilot challenge → See how it works

1 real task in 1 ranked shortlist out Pune, India · est. 2024

challenge / rate_limiter.py timed · 90:00

# Task: fix the token-bucket refill under burstdef allow(bucket, now):    elapsed = now - bucket.last    refill  = elapsed * bucket.rate    bucket.tokens = min(bucket.cap, bucket.tokens + refill)    bucket.last = now    if bucket.tokens >= 1:        bucket.tokens -= 1        return True    return False

pytest · 6 checksruntime 0.42s

✓ handles steady traffic pass

✓ caps burst at bucket size pass

✓ refills fractionally over time pass

✓ rejects when empty pass

✓ monotonic clock edge case pass

✓ concurrent access safe pass

6 / 6 passing · clean code + edge casesscore 94

the loop

Three ways to measure real skill

Same platform, different levers. Pick the format that gives you the depth of signal you need for the role.

01 / challenge

Coding challenges

Timed, real tasks for volume screening. Everyone gets the same brief and clock — take-home for depth or live for pressure.

02 / hackathon

Hackathons

Weekend builds where you find who can ship something real under pressure — and collaborate while doing it.

03 / real task

Real-world tasks

A bug from your backlog, an API to design, a pipeline to fix. The part a resume can't fake — how they reason on a messy spec.

open req → shortlist

From one task to a ranked shortlist

Task

Bring a real problem from your codebase, or pick one from our library matched to the role.

Run

Candidates get the brief and a fair clock. Everyone works the same problem, so results compare.

Score

Automated tests handle correctness; a fixed rubric scores quality, edge cases and reasoning.

Leaderboard

A ranked view with the work attached. Click into any submission and see exactly why they placed.

objective, not cold

Scored on a rubric written before the first submission

Candidates aren't scored against each other's vibe — they're scored against a fixed standard, then ranked. The reasons are visible and auditable.

Correctness40%

Code quality25%

Edge cases20%

Reasoning15%

leaderboard

Ranked by output, not origin story

The shortlist stops looking like a school reunion and starts looking like your best candidates.

01candidate_7f2
python · 90:006/6 tests94

02candidate_a1c
go · 74:126/6 tests91

03candidate_e90
typescript · 88:405/6 tests83

04candidate_44b
rust · 90:005/6 tests79

for employers

Why skill-first funnels win

Teams burn weeks screening for signals that don't predict on-the-job performance. Change what you measure, and the whole conversation upgrades.

Skill over resume

Put everyone in front of the same real task and rank by the output. The paper and the skill are two different things — measure the one that survives day one.

Auto-scoring at scale

Automated test suites do the objective pass/fail; a structured review scores the rest. The leaderboard does your first cut, so engineers review the top ten, not 200 resumes.

Faster time-to-hire

Time from open req to a confident offer shrinks from weeks to days. The interview stops being a re-screen and becomes a real conversation.

Defensible & blind

Everyone graded on the same lines under the same clock. A hiring manager can override the rank — but they have to look at the code to do it.