Databricks interviews

Work in the tool while you explain your decisions.

Choose a role and level. Each interview has a new task, follow-up questions and a report that checks your work and your explanation.

Take the assessment Choose an interview

Speak or type. Sign in to save your results across devices.

Timed assessment

12 practical questions
50 minutes
Mid level

Questions are drawn from a bank of 36. Your report includes a skill breakdown and lessons to practice next.

Start assessment

Practice interviews

Junior tasks cover the fundamentals. Mid and senior tasks add debugging, changing requirements and review decisions.

Position

10 interviews

Analytics engineer
Junior
20 min

Juniper Service: SQL population checks

Juniper Service needs district, intake_rows (all intake versions), and signed_credit (SUM of credit) from main.bronze.screen_501. Include the NULL district group, zero credit and negative corrections. Count the raw versions separately.

Start interview

Skills and rubricSQL population checks

Builds the stated sql population checks result and identifies the source population.
Uses the supplied missing-value or boundary record to test the result.
Explains the purpose of each selected output field.

Data engineer
Junior
20 min

Alder Parcel: PySpark missing-value transformation

In PySpark, read main.bronze.screen_502 and return ticket_id, credit_known (true for every non-NULL credit), and display_credit with NULL replaced by zero. Keep every version. Show the resulting DataFrame.

Start interview

Skills and rubricPySpark missing-value transformation

Builds the stated pyspark missing-value transformation result and identifies the source population.
Uses the supplied missing-value or boundary record to test the result.
Explains the purpose of each selected output field.

Data engineer
Junior
20 min

Wren Repairs: Deterministic source versions

Wren Repairs has duplicate versions in main.bronze.screen_503. Return every field for one current row per ticket_id, choosing latest arrived_at (NULLS LAST). Exact duplicates may collapse. Select versions by their update order.

Start interview

Skills and rubricDeterministic source versions

Builds the stated deterministic source versions result and identifies the source population.
Uses the supplied missing-value or boundary record to test the result.
Explains the purpose of each selected output field.

Analytics engineer
Mid
30 min

Juniper Service: Join cardinality

For every account in main.bronze.accounts_504, return account_id, current_tickets and signed_credit (zero if no known credit). Join only one current row per ticket from main.bronze.screen_504, using latest arrived_at then highest credit, both NULLS LAST. Exclude unassigned tickets from these account totals.

Start interview

Skills and rubricJoin cardinality

Handles join-cardinality without constants tied to visible rows.
Applies the changed requirement and checks replay or boundary behavior.
Explains a realistic wrong approach using the supplied counterexample.

Analytics engineer
Mid
30 min

Alder Parcel: Ranking with ties

Alder Parcel asks for ticket_id, district and credit for one highest-credit current ticket in each district. First choose current versions by arrived_at then credit descending, both NULLS LAST. Exclude NULL credit, preserve negative credit when it's a district's highest value.

Start interview

Skills and rubricRanking with ties

Handles window-ties without constants tied to visible rows.
Applies the changed requirement and checks replay or boundary behavior.
Explains a realistic wrong approach using the supplied counterexample.

Analytics engineer
Mid
30 min

Wren Repairs: Complete timestamp intervals

Return service_day, tickets and signed_credit from main.bronze.screen_506 for the complete UTC service interval May 1 through May 7, 2025. event_at is already normalized to UTC. Group by event date, keep signed credits and all versions, and exclude missing event_at.

Start interview

Skills and rubricComplete timestamp intervals

Handles timestamp-boundary without constants tied to visible rows.
Applies the changed requirement and checks replay or boundary behavior.
Explains a realistic wrong approach using the supplied counterexample.

Data engineer
Mid
30 min

Juniper Service: Delta keyed upsert

Apply main.bronze.corrections_507 to main.silver.screen_507, matching ticket_id. Choose one incoming version per key using arrived_at then credit descending, both NULLS LAST. Replace matched rows and insert new keys.

Start interview

Skills and rubricDelta keyed upsert

Handles delta-upsert without constants tied to visible rows.
Applies the changed requirement and checks replay or boundary behavior.
Explains a realistic wrong approach using the supplied counterexample.

Data platform engineer
Senior
40 min

Alder Parcel: Delta state recovery

A mistaken update changed every credit in main.silver.screen_508. Recover exactly the table's original creation state at version 0 using Delta recovery. Preserve the full schema and NULLs. Inspect retained history before explaining the recovery.

Start interview

Skills and rubricDelta state recovery

Defends the delta-recovery contract and its downstream effect.
Keeps correctness evidence separate from performance assumptions.
Names a release or refresh check and a responsible owner.

Data platform engineer
Senior
40 min

Wren Repairs: Pipeline expectations and evidence

Publish main.silver.screen_509 as a streaming table from main.bronze.screen_509. Name the quality expectation known_positive, require credit > 0, and DROP ROW on violations. Keep every valid input version. Preserve the named expectation's pass and failure metrics.

Start interview

Skills and rubricPipeline expectations and evidence

Defends the pipeline-quality contract and its downstream effect.
Keeps correctness evidence separate from performance assumptions.
Names a release or refresh check and a responsible owner.

Data platform engineer
Senior
40 min

Juniper Service: Optimization review

The platform owner approved OPTIMIZE main.silver.screen_510 with ZORDER BY account_id for an existing workload investigation. Execute that operation while preserving all rows. Retain history as evidence. The local engine records the request without executing distributed file compaction; make no speed or cost claim.

Start interview

Skills and rubricOptimization review

Defends the performance-review contract and its downstream effect.
Keeps correctness evidence separate from performance assumptions.
Names a release or refresh check and a responsible owner.

Choose your character

Databricks interviews

Timed assessment

Practice interviews

Juniper Service: SQL population checks

Alder Parcel: PySpark missing-value transformation

Wren Repairs: Deterministic source versions

Juniper Service: Join cardinality

Alder Parcel: Ranking with ties

Wren Repairs: Complete timestamp intervals

Juniper Service: Delta keyed upsert

Alder Parcel: Delta state recovery

Wren Repairs: Pipeline expectations and evidence

Juniper Service: Optimization review

Interview Hub access

Your Work Gym account