Query like a data engineer.

SELECT, WHERE, NULL buckets, CASE bands, and your first GROUP BY sanity checks.

Inner and left joins, anti-joins with IS NULL, EXISTS semi-joins, and referential checks.

GROUP BY, HAVING, conditional aggregation, and decimal-safe ratios per country and buyer.

Named CTE layers, scalar and correlated subqueries, derived tables, and readable narration.

ROW_NUMBER dedupe, ranks, LAG gaps, running totals, and a first cohort retention slice.

Event time vs arrival time, time zones, JSON extraction, array expansion, and late-data audits.

Incremental load classification, merge deltas, SCD2 boundaries, backfill plans, reconciliation.

Duplicate keys, completeness, referential integrity, total reconciliation, and audit summaries.

Partition pruning, full-scan detection, scan cost per analyst, hot-partition skew, wide scans.

Capstone interview scenarios — output contracts, fanout-safe revenue, funnels, SCD2 checks, cost triage.