Assessments · Coding
A real editor, a real work sample.
Candidates write real code in a real editor — not a quiz about code. For Python and JavaScript, it's executed for real in an isolated sandbox against visible and hidden test cases. Other languages are saved exactly as written, for human review.
10# Rebalance workloads across nodes 11def rebalance(nodes, threshold=0.8): 12 hot = [n for n in nodes if n.load > threshold] 13 for node in sorted(hot, key=lambda n: -n.load): 14 target = min(nodes, key=lambda n: n.load) 15 node.migrate_to(target) 16 return nodes
Nothing you write is ever lost
Run in an isolated sandbox — no network access, no access to any other data. Other languages can still be written; they're saved as a work sample instead of executed.
Every submission is saved as a durable record the moment it's sent — nothing is silently lost.
Candidates can run their solution against the visible test cases as often as they like; submitting also scores it against hidden ones they can't see.
Candidate experience
Autosave, resume,
no ambushes
Every keystroke is saved. A dropped connection pauses safely — work is never lost. Timers warn at half-time and five minutes, and back-navigation never destroys progress.
How integrity worksQuestion 3 of 4 · every keystroke is saved