Your question is SQL Query Interpretation. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
An engineer on Scale's labeling-ops team sends you this query and tells you it's what powers the "September completions per annotator" number on your weekly PM dashboard.
-- labeling_tasks: task_id, annotator_id, completed_at, status
-- task_reviews: review_id, task_id, reviewer_id, review_status
SELECT
t.annotator_id,
COUNT(t.task_id) AS tasks_completed,
AVG(CASE WHEN r.review_status = 'approved' THEN 1 ELSE 0 END) AS approval_rate
FROM labeling_tasks t
JOIN task_reviews r ON t.task_id = r.task_id
WHERE t.completed_at >= '2026-09-01'
GROUP BY t.annotator_id;
A task can go through more than one review. Before you repeat this number to a stakeholder, explain exactly what each row of this result represents, whether it actually matches "completions per annotator in September," and what you'd change.