A data engineer needs to pull customer records from a legacy CRM that exposes only a REST API. The API returns JSON, but the CRM's API has a rate limit of 100 requests per minute. The engineer must extract 50,000 records and store them in a data lake. The engineer writes a Python script that loops through pages of 100 records each. After running for a few minutes, the script starts receiving HTTP 429 responses. What should the engineer do to complete the extraction reliably?

Increase the page size to 1,000 records per request

Switch to a streaming platform such as Kafka to ingest the data

Implement exponential backoff with retries on 429 responses

Run multiple parallel scripts to fetch different pages simultaneously

A data science team is acquiring data from a public government API. The API documentation states that fields may be missing or contain placeholder values like 'N/A'. The team needs to decide whether the data is suitable for training a model. What should they do first?

Replace all placeholders with the mean value

Train a model and see if it performs well

Assume the data is fine because it is from a government source

Perform exploratory data analysis to quantify missing values and placeholders

A machine learning team is training models on a dataset that is updated weekly. They need to ensure that any model trained can be traced back to the exact data version used. Which practice should they implement?

Save the dataset in a shared folder with a descriptive name

Keep only the latest version of the dataset in the storage system

Rely on the file modification timestamp to identify the data version

Store a hash of the dataset and record it in the model training metadata

A machine learning team is developing a recommendation system. They use a dataset that is updated daily from multiple sources. The team needs to be able to reproduce any model trained in the past. Which combination of practices should they implement?

Store the dataset in a database and overwrite it daily

Version the dataset daily and store the version ID in the model training metadata

Keep only the latest dataset and rely on the model's training date

Use a file naming convention with the date and hope it is enough

A retail company needs to combine sales data from a cloud-based CRM API, a legacy on-premises database, and a third-party data provider that delivers daily files. The combined data will feed a demand forecasting model. Which acquisition strategy is most appropriate?

Use the CRM API to pull data from all three sources into one unified dataset.

Build separate ingestion jobs for each source and land the raw data in a common staging area before merging.

Manually download the legacy database and third-party files into spreadsheets, then upload them to the cloud.

Ask the third-party provider to send data directly to the legacy database.

GIAC Machine Learning Engineer

Domain 1Objective 1

Data Acquisition GMLE Practice Questions (Page 4)

Part of the Foundational Data Science and Statistics domain, which makes up ~41% of our current practice bank. GIAC (SANS) does not publish an official question count, but from its 180-minute exam (~70–120 total, ~29–49 in this domain), expect 7–12 from this objective — we provide 39 practice questions to prepare you well beyond it. (estimate)

39questions here

8free pages

5concepts

Questions 16–20

16application · medium
A data engineer needs to pull customer records from a legacy CRM that exposes only a REST API. The API returns JSON, but the CRM's API has a rate limit of 100 requests per minute. The engineer must extract 50,000 records and store them in a data lake. The engineer writes a Python script that loops through pages of 100 records each. After running for a few minutes, the script starts receiving HTTP 429 responses. What should the engineer do to complete the extraction reliably?
Select an answer first
17expert · hard
A data science team is acquiring data from a public government API. The API documentation states that fields may be missing or contain placeholder values like 'N/A'. The team needs to decide whether the data is suitable for training a model. What should they do first?
Select an answer first
18application · medium
A machine learning team is training models on a dataset that is updated weekly. They need to ensure that any model trained can be traced back to the exact data version used. Which practice should they implement?
Select an answer first
19expert · hard
A machine learning team is developing a recommendation system. They use a dataset that is updated daily from multiple sources. The team needs to be able to reproduce any model trained in the past. Which combination of practices should they implement?
Select an answer first
20application · medium
A retail company needs to combine sales data from a cloud-based CRM API, a legacy on-premises database, and a third-party data provider that delivers daily files. The combined data will feed a demand forecasting model. Which acquisition strategy is most appropriate?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to GMLE

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by GIAC (SANS). “GMLE” is a trademark of its owner, used for identification only.