Compare each row to its neighbor without a self-join.

Aggregate over a sliding window of rows with a frame clause.

Walk self-referencing trees like org charts and category paths.

Populate dimensions with surrogate keys, then load a fact that references them.

Overwrite a changed attribute in place with no history.

Preserve history by expiring old rows and inserting new versions.

Denormalize a wide star dimension into a snowflake sub-dimension, then pay the extra join to query it.

Classify fact tables by grain and use measure additivity to decide which sums are valid: correctly roll up a semi-additive balance and recompute a non-additive ratio from additive parts.

Model conformed, role-playing, degenerate, and junk dimensions, plus factless facts, the way data-engineering interviewers ask about them.

Keep exactly one row per business key from a dirty source.

Make a re-run produce the same result: no duplicated rows.

Encode expectations as tests that fail before bad data spreads.

Read a query plan and turn a full scan into an index seek.

Combine dedup, SCD2, idempotency, quality, and perf into one production-grade loader.