Real engineering problems. Better engineering judgment.
Practise investigating broken deployments, reading logs, finding faulty code, and defending your engineering decisions.
How Trace works
Trace replaces synthetic interview puzzles with real production failure modes.
Investigate
Examine code changes, logs, and system context.
Inspect realistic code diffs, server logs, and user reports without artificial clues. Click directly on diff lines to test your hypothesis against deterministic checks against curated incident answer keys.
Decide and defend
Choose how to contain a failure, identify a fault, and explain a permanent fix.
Lock in immediate mitigation, write your permanent fix, and answer tailored trade-off questions challenging concurrency, caching, or backward-compatibility.
Review — Learn
See what code verified, what AI interpreted, and how your reasoning compares with a curated reference.
Visibly separate deterministic code truth from AI evaluation. Rubric criteria cite your exact words verbatim, backed by production-ready reference solutions.
Ready to investigate your first incident?
Start with a beginner-friendly API breakage or dive into a distributed concurrency race condition.