01. A bank is consolidating three regional data warehouses and a data lake. Someone must define the target landscape: which platforms to keep, how data flows between them and which modeling standards every team follows.
Which role is responsible for this?
a) Data scientist
b) Data analyst
c) Data engineer
d) Data architect
02. A bank runs its interest calculations as a batch job overnight instead of computing them throughout the business day.
Why does this approach use the bank's computing resources efficiently?
a) Accumulated records are processed together in one run at a quiet time with spare capacity
b) The job runs continuously alongside daytime workloads, so servers never sit idle between tasks
c) Each transaction is processed the instant it occurs, so no data accumulates
d) Results become available within milliseconds, so fewer reports have to be rerun
03. A company is moving its order-processing database to the cloud. The database engine needs a durable raw volume that it formats with its own file system and that gives consistently low latency for many small random reads and writes.
Which storage type fits this requirement?
a) Object storage
b) In-memory cache storage
c) Block storage
d) File storage
04. A data engineer is documenting the stages of a new machine learning pipeline for a demand-forecasting model.
Which sequence places the stages in the correct order?
a) Data collection, data preparation, model training, feature engineering, model evaluation
b) Data preparation, data collection, feature engineering, model evaluation, model training
c) Data collection, feature engineering, data preparation, model training, model evaluation
d) Data collection, data preparation, feature engineering, model training, model evaluation
05. A database designer is explaining to a new team member why several indexes have been added to the Orders table.
Which statement best describes the role of an index?
a) It supplies a value for a column when an insert leaves it out
b) It enforces that every order references a valid customer record
c) It lets queries locate rows without scanning the whole table
d) It records which other tables are related to the orders table
06. A pipeline has three stages. Ingestion and loading each keep up with the incoming data easily, but the transformation stage runs at full capacity and records queue up in front of it.
Which change raises the pipeline's overall throughput?
a) Adding capacity to ingestion and loading together
b) Adding capacity to the transformation stage
c) Adding capacity to the ingestion stage
d) Adding capacity to the loading stage
07. A seismic survey company records several petabytes of raw sensor traces each year. Researchers may reprocess any survey years later, but most traces are never read again. Finance wants the lowest storage cost that still keeps every trace in its original form.
Where should the traces be kept?
a) In a relational data warehouse, modeled into tables as they are loaded
b) In a separate data mart for each research team, holding only its own surveys
c) In a data lake that keeps the raw files on low-cost storage
d) In an in-memory key-value store, so each trace is fetched instantly by its ID
08. A team adds indexes on five columns of an orders table. Soon afterwards, the nightly bulk load of new orders takes noticeably longer.
What explains the slower load?
a) Indexes lock the table against reads, so the reporting queries block the load
b) Each insert must also write an entry into every index on the table
c) Indexes replace the primary key, so duplicates must be checked by a full scan
d) Indexes compress the table, so each insert must first decompress the existing rows on disk
09. A team is building a new customer-analytics platform. Its plan is to launch quickly and add privacy controls once the platform is in use.
Which change to the plan applies privacy by design?
a) Define minimization, pseudonymization and access rules as requirements before building
b) Publish a privacy notice, train the support team and log customer complaints
c) Encrypt the platform's data only if a regulator or customer later asks for it
d) Launch as planned and commission a privacy review after the first quarter
10. An online bank wants to flag a login session as suspicious when the same session attempts several failed transfers to new payees within a short period. Each event in the stream describes only one action.
Why does this detection require stateful stream processing?
a) Because each event can be judged on its own, without reference to earlier ones
b) Because the events must be written to the warehouse before any processing begins
c) Because earlier events in the same session must be remembered to judge the latest one
d) Because the results are only needed later, in a monthly fraud report for the risk team