When do you keep Airflow instead of moving to Dagster?

A large heterogeneous ops estate, vendor sensors, and a team already staffed on Airflow. Move when the DAG is secretly a data graph and assets would delete half the glue. Fashion is not a reason.

Someone DELETED gold.orders. What do you do in the first ten minutes?

Query History for the last good timestamp. SELECT count(*) … AT (TIMESTAMP => …). CLONE to a _fix table and diff. Do not UNDROP as the first move if writers are live — you can race a reload. Time Travel is a restore window, not a backup policy. Fail-safe is Snowflake-only and not queryable.

Zero-copy clone of prod for a destructive test. What do you warn finance?

Clone is cheap at t0. Storage grows as either side writes. Time Travel on the clone is not a backup of prod. Drop the clone when the test ends.

Consultancy · Interview pack

Deloitte data interview pack

13 written questions with answers, the loop shape people report for Deloitte, and the stacks to practise. Built from DataLane's original question bank and publicly discussed interview themes, not leaked exams. Last editorial update 2026-06-01.

What is in this pack

Questions by topic

Orchestration1
Warehouse6
Cloud6

4 easy · 7 medium · 2 hard

Rounds (1)

Case

Themes: Diagrams, Medallion, Client constraints

Free with a DataLane account

The first 3 questions are below. Sign in (free, no plan needed) to read all 13 and download the pack as a PDF.

Free preview

1. Orchestration · medium · Dagster
When do you keep Airflow instead of moving to Dagster?
A large heterogeneous ops estate, vendor sensors, and a team already staffed on Airflow. Move when the DAG is secretly a data graph and assets would delete half the glue. Fashion is not a reason.
2. Warehouse · medium · Snowflake
Someone DELETED gold.orders. What do you do in the first ten minutes?
Query History for the last good timestamp. SELECT count(*) … AT (TIMESTAMP => …). CLONE to a _fix table and diff. Do not UNDROP as the first move if writers are live — you can race a reload. Time Travel is a restore window, not a backup policy. Fail-safe is Snowflake-only and not queryable.
3. Warehouse · medium · Snowflake
Zero-copy clone of prod for a destructive test. What do you warn finance?
Clone is cheap at t0. Storage grows as either side writes. Time Travel on the clone is not a backup of prod. Drop the clone when the test ends.

10 more questions are in the full pack.

The Deloitte loop

Case
Draw bronze/silver/gold for a client who already has Synapse and wants Fabric. What you would not migrate first.

Orchestration (1)

medium · Dagster
When do you keep Airflow instead of moving to Dagster?
A large heterogeneous ops estate, vendor sensors, and a team already staffed on Airflow. Move when the DAG is secretly a data graph and assets would delete half the glue. Fashion is not a reason.

Warehouse (6)

medium · Snowflake
Someone DELETED gold.orders. What do you do in the first ten minutes?
Query History for the last good timestamp. SELECT count(*) … AT (TIMESTAMP => …). CLONE to a _fix table and diff. Do not UNDROP as the first move if writers are live — you can race a reload. Time Travel is a restore window, not a backup policy. Fail-safe is Snowflake-only and not queryable.
medium · Snowflake
Zero-copy clone of prod for a destructive test. What do you warn finance?
Clone is cheap at t0. Storage grows as either side writes. Time Travel on the clone is not a backup of prod. Drop the clone when the test ends.
easy · Snowflake
Cortex Analyst returns doubled revenue. First check?
The semantic model grain and joins — same fan-out you would debug in SQL. A bigger warehouse will not fix a bad metric definition.
hard · Snowflake
Secure view vs masking policy vs row-access policy. One sentence each.
Secure view hides the SQL and blocks some optimizations. Masking rewrites a column per role. Row-access filters rows. Do not stack all three “for security theater.”
easy · Databricks
Unity Catalog vs Hive metastore. Why do interviewers care in 2026?
UC is the governance plane: identities, row filters, lineage, a 3-level namespace. HMS is the legacy catalog. “We still have two catalogs” is a real migration answer. Do not claim ACLs on a path-based table you never registered.
easy · Databricks
Unity Catalog volume vs a DBFS path in a notebook. What do you refuse to merge?
Hard-coded dbfs:/ in production jobs. Volumes + UC grants are the paved road. DBFS mounts with a personal token are how you leak prod.

Cloud (6)

medium · AWS
Redshift DISTKEY / SORTKEY (or AUTO). What mistake shows up as a broadcast of a 2 TB table?
A join on a column that is not the dist key, so every slice shuffles everything. Even in AUTO, a bad mental model (SELECT * fact JOIN huge dim ON a non-selective key) will not be saved by a wizard. Explain slice-local joins in one sentence.
hard · AWS
Lake Formation LF-Tags vs bucket policy. When is LF worth the ticket?
When many accounts/roles need column- and table-level grants on a shared catalog. A single team with two buckets can stay on IAM + bucket policy. LF is not a substitute for not putting PII in the key name.
easy · Azure
Self-hosted IR versus Azure IR. When is on-prem the only answer?
Azure IR talks to public PaaS. A self-hosted IR sits inside the VNet / on-prem so you can reach SQL Server that has no public endpoint. “We opened 1433 to the internet so we could use Azure IR” is the wrong answer.
medium · Azure
Purview vs “the pipeline wiki.”
A catalog with owners and classifications beats a Confluence page that rot. If you cannot name the producer, you cannot page.
medium · Azure
Fabric Direct Lake vs Import. When is Direct Lake the wrong choice?
When the lake table is not Delta/Iceberg-shaped for the semantic model, or you need a frozen extract for a board meeting. Direct Lake still needs a clean gold grain. Import is not “legacy” if the model is 400 MB.
medium · Azure
Purview scan never sees your gold tables. First three causes?
Wrong collection/scope, the identity cannot list the catalog, or tables live in a workspace Purview is not connected to. A wiki page is not a scan. Fix the connection before buying another catalog.