Drill the part you are weakest on.
The Daily Brick gives everyone the same five questions. Practice lets you choose: one topic at one difficulty, or the mixed set for your role. Same question bank, same rule: every code question ran on Spark before it shipped.
By role
A mixed set drawn across the topics that role runs into.
Delta semantics, pipeline correctness, and the join and null rules behind most bad numbers.
SQL semantics, window functions, and aggregation edge cases.
Table mechanics and the storage behaviour that shows up as a cost or a corruption question.
By topic
Ten questions at the difficulty you pick.
- Core: semantics you should have
- Edge cases: the ones you have not hit yet
- Subtle: the ones that only show up in production
Versions, time travel, MERGE, and what a commit actually rewrites.
DataFrame semantics, nulls, joins, and lazy evaluation.
Window functions, aggregation, and the null rules that decide a result.
The transformations where correct and merely quiet diverge.
Constraints, dedup, and the silent row loss nobody notices.