SQL

Read a table and its grain, shape a query, select and filter deliberately, handle types and NULLs, count safely, and inspect results like an engineer.

Join types, grain, cardinality, duplicate explosions, and the habit of proving why rows multiplied.

Grouped metrics, HAVING filters, distinct counts, conditional aggregation, and trustworthy metric definitions.

Layered query design, debuggable intermediate results, and SQL that is easy to narrate in interviews.

Ranking, offsets, running totals, deduplication, retention, cohorts, and event sequence analysis.

Timestamp logic, time zones, event tables, JSON, arrays, late data, and ordering tradeoffs.

UNION, INTERSECT and EXCEPT; pivoting with conditional aggregation, unpivoting with UNION ALL, and subtotal rows you cannot misread.

Deriving episodes from events: unbroken runs, streaks, sessionisation with a threshold, ordered funnels, and honest cohort retention.

Incremental loads, MERGE patterns, upserts, slowly changing dimensions, snapshots, and backfills.

Duplicate checks, null checks, referential tests, reconciliation, anomaly queries, and audit evidence.

What you are renting when you run SQL against a cloud warehouse, and which of the three layers — storage, compute, or metadata — to blame for a bill, a slow dashboard, or a load that did nothing.

Partitions, clustering, pruning, query plans, scan cost, and warehouse-aware optimization habits.

Clarifying assumptions, narrating query logic, debugging live, and explaining senior-level tradeoffs.

The transactional side of performance: B-trees, selectivity, sargable predicates, composite key order, index write cost, and why none of it exists on a warehouse.