Each live-coding task comes with specific constraints designed to test targeted competencies. We evaluate whether the solution works, follows best practices, avoids anti-patterns, and stays within the stated constraints. Scoring is grade-aware: Juniors get credit for correct-but-straightforward solutions. Mids need to show understanding of tradeoffs. Seniors are expected to handle edge cases thoughtfully - without over-engineering. Plus, our red-flag matrix auto-caps scores (~30–40 range) if a candidate confuses the stack or writes code that contradicts how the technology actually behaves - even if the rest looks passable