Platform · Challenges
Real engineering challenges, on production-shaped code.
Four categories, three difficulties, one scoring rubric. Browse the catalog, or hop into a session and let the system route you to the next challenge that stretches your weakest dimension.
Categories
Pick a domain. The system picks the next challenge.
Debugging
Production-shaped bugs across Node, Python, and Go. Memory leaks, race conditions, silent data loss, transaction drift, intermittent timeouts.
Security
OWASP-class issues, prompt-injection-as-RCE, SSRF in tool-using agents, auth boundary drift, leaky logs. Includes red-team rounds against AI-generated code.
Refactoring
Take a 400-line component, refactor with an AI pair, defend your decomposition. Pattern fluency across services, hooks, and module boundaries.
Feature Build
Real ticket against a real backlog. You ship the feature: tests, migrations, telemetry, PR description. Graded end-to-end.
Difficulty
Three difficulties, one rubric.
Time budgets are guidance, not constraints. We measure outcomes — correctness, regression rate, defensibility — across all bands.
Easy
Warm-ups. A single bug, a tight scope, a defined acceptance criterion. Useful for daily practice or onboarding new hires.
Medium
The core track. Multi-file scope, ambiguous spec, real debugging or refactoring decisions. The bulk of certification practice lives here.
Hard
Interview and certification class. Cross-system context, latent failure modes, AI-generated traps. Graded by the full rubric, panel-reviewable.