support@Free-Braindumps.com
Register Login
Cisco
CompTIA
ISC
EC-Council
EXIN
Fortinet
ITIL
Juniper
Microsoft
VMware
Avaya
SAP
Google
NVIDIA
Salesforce
Amazon
Palo Alto Networks
ISC2
Splunk
ServiceNow
All Vendors
  1. Home
  2. Exams
  3. Google
  4. PROFESSIONAL MACHINE LEARNING ENGINEER

Google PROFESSIONAL MACHINE LEARNING ENGINEER Exam Actual Questions
Professional Machine Learning Engineer (Page 4 )

Updated On: 12-Jul-2026
Page 4 of 44
PDF with 339 Questions

You are building a real-time prediction engine that streams files which may contain Personally Identifiable Information (PII) to Google Cloud. You want to use the Cloud Data Loss Prevention (DLP) API to scan the files.

How should you ensure that the PII is not accessible by unauthorized individuals?

  1. Stream all files to Google Cloud, and then write the data to BigQuery. Periodically conduct a bulk scan of the table using the DLP API.
  2. Stream all files to Google Cloud, and write batches of the data to BigQuery. While the data is being written to BigQuery, conduct a bulk scan of the data using the DLP API.
  3. Create two buckets of data: Sensitive and Non-sensitive. Write all data to the Non-sensitive bucket.
    Periodically conduct a bulk scan of that bucket using the DLP API, and move the sensitive data to the Sensitive bucket.
  4. Create three buckets of data: Quarantine, Sensitive, and Non-sensitive. Write all data to the Quarantine bucket. Periodically conduct a bulk scan of that bucket using the DLP API, and move the data to either the Sensitive or Non-Sensitive bucket.

Answer(s): D

Explanation:

Option D is correct because it implements a containment workflow: initial quarantine of all streams, periodic DLP scanning, and controlled movement to separate buckets based on sensitivity, minimizing unauthorized access to PII during processing. This aligns with least-privilege and data lifecycle practices in GCP, and ensures only reviewed data is promoted to sensitive storage.
A) Incorrect — writing to BigQuery and bulk scanning the table does not quarantine or prevent access during ingestion; potential exposure before classification.
B) Incorrect — streaming into BigQuery with interim DLP scan does not isolate unanalyzed data; access control and data at-rest protection are not guaranteed during ingestion.
C) Incorrect — simply separating into Non-sensitive and Sensitive without a quarantine and explicit workflow risks premature access and incomplete visibility before classification.



You work for a large hotel chain and have been asked to assist the marketing team in gathering predictions for a targeted marketing strategy. You need to make predictions about user lifetime value (LTV) over the next 20 days so that marketing can be adjusted accordingly. The customer dataset is in BigQuery, and you are preparing the tabular data for training with AutoML Tables. This data has a time signal that is spread across multiple columns. How should you ensure that AutoML fits the best model to your data?

  1. Manually combine all columns that contain a time signal into an array. AIlow AutoML to interpret this array appropriately. Choose an automatic data split across the training, validation, and testing sets.
  2. Submit the data for training without performing any manual transformations. AIlow AutoML to handle the appropriate transformations. Choose an automatic data split across the training, validation, and testing sets.
  3. Submit the data for training without performing any manual transformations, and indicate an appropriate column as the Time column. AIlow AutoML to split your data based on the time signal provided, and reserve the more recent data for the validation and testing sets.
  4. Submit the data for training without performing any manual transformations. Use the columns that have a time signal to manually split your data. Ensure that the data in your validation set is from 30 days after the data in your training set and that the data in your testing sets from 30 days after your validation set.

Answer(s): D

Explanation:

Option D is correct because AutoML Tables can utilize a time-based split when no manual transformation is done, ensuring forward-looking evaluation by placing training, validation, and test data in chronological order (e.g., 30-day gaps). This preserves temporal integrity for LTV forecasts and prevents leakage.
A) Incorrect — manually combining time signals into an array is not a supported or recommended AutoML convention for temporal modeling; it risks misinterpretation and data leakage.
B) Incorrect — AutoML Tables needs a temporal split strategy; relying on automatic splits can leak future information in time-series-like data.
C) Incorrect — while indicating a Time column helps, AutoML does not perform a time-based split automatically unless explicitly configured; the recommended approach maintains forward-looking evaluation through explicit temporal partitioning.



You have written unit tests for a Kubeflow Pipeline that require custom libraries. You want to automate the execution of unit tests with each new push to your development branch in Cloud Source Repositories. What should you do?

  1. Write a script that sequentially performs the push to your development branch and executes the unit tests on Cloud Run.
  2. Using Cloud Build, set an automated trigger to execute the unit tests when changes are pushed to your development branch.
  3. Set up a Cloud Logging sink to a Pub/Sub topic that captures interactions with Cloud Source Repositories.
    Configure a Pub/Sub trigger for Cloud Run, and execute the unit tests on Cloud Run.
  4. Set up a Cloud Logging sink to a Pub/Sub topic that captures interactions with Cloud Source Repositories.
    Execute the unit tests using a Cloud Function that is triggered when messages are sent to the Pub/Sub topic.

Answer(s): B

Explanation:

Option B is correct because Cloud Build can be triggered automatically by pushes to Cloud Source Repositories, enabling CI for unit tests on each development branch update. It integrates with Cloud Source Repositories and supports running test steps in isolated build environments.
A) Incorrect — manually scripting pushes and Cloud Run executions defeats the purpose of automated CI/CD and lacks scalable, repeatable trigger-based execution.
C) Incorrect — Cloud Logging sinks to Pub/Sub capture logs, not trigger CI workflows for code changes.
D) Incorrect — Duplicate of C; same reasoning applies, not a CI trigger mechanism.



You are training an LSTM-based model on AI Platform to summarize text using the following job submission script:

gcloud ai-platform jobs submit training $JOB_NAME \
--package-path $TRAINER_PACKAGE_PATH \
--module-name $MAIN_TRAINER_MODULE \
--job-dir $JOB_DIR \
--region $REGION \
--scale-tier basic \
-- \
--epochs 20 \
--batch_size=32 \
--learning_rate=0.001 \

You want to ensure that training time is minimized without significantly compromising the accuracy of your model. What should you do?

  1. Modify the `epochs' parameter.
  2. Modify the `scale-tier' parameter.
  3. Modify the `batch size' parameter.
  4. Modify the `learning rate' parameter.

Answer(s): B

Explanation:

Option B is correct because scaling tier affects allocated resources (CPU/GPU/TPU), which directly reduces training time on AI Platform without changing model architecture; higher tiers provide more parallelism and faster wall-clock time. A) Modifying epochs changes training duration but not speed per epoch and can hurt accuracy if not tuned carefully. C) Batch size can impact memory and throughput but may require learning-rate tuning and can affect convergence; not primarily a speed-up knob. D) Learning rate affects convergence, not training time; incorrect changes can slow or destabilize training rather than shorten wall-clock time.



You have deployed multiple versions of an image classification model on AI Platform. You want to monitor the performance of the model versions over time. How should you perform this comparison?

  1. Compare the loss performance for each model on a held-out dataset.
  2. Compare the loss performance for each model on the validation data.
  3. Compare the receiver operating characteristic (ROC) curve for each model using the What-If Tool.
  4. Compare the mean average precision across the models using the Continuous Evaluation feature.

Answer(s): D

Explanation:

Option D is correct because Continuous Evaluation in AI Platform allows tracking and comparing metrics like mean average precision (MAP) across deployed model versions over time, enabling principled performance progression assessment for image classification.
A) Incorrect — loss on a held-out dataset may not reflect real-world ranking or detection quality; MAP is a more appropriate summary for multi-label/classification ranking.
B) Incorrect — validation data is typically static; continuous deployment monitoring requires ongoing evaluation, not a one-off validation split.
C) Incorrect — ROC curves are informative for binary classifiers; multilabel image classification often relies on MAP or precision-recall metrics, and What-If Tool isn’t the standard continuous cross-version comparator.



You trained a text classification model. You have the following SignatureDefs:



You started a TensorFlow-serving component server and tried to send an HTTP request to get a prediction using:

headers = {"content-type": "application/json"}
json_response = requests.post('http: //localhost:8501/v1/models/ text_model:predict', data=data, headers=headers)

What is the correct way to write the predict request?

  1. data = json.dumps({"signature_name": "seving_default", "instances" [[`ab', `bc', `cd']]})
  2. data = json.dumps({"signature_name": "serving_default", "instances" [[`a', `b', `c', `d', `e', `f']]})
  3. data = json.dumps({"signature_name": "serving_default", "instances" [[`a', `b', `c'], [`d', `e', `f']]})
  4. data = json.dumps({"signature_name": "serving_default", "instances" [[`a', `b'], [`c', `d'], [`e', `f']]})

Answer(s): D

Explanation:

Option D is correct because TensorFlow Serving REST predict requests require a JSON payload with signature_name set to serving_default and instances as a list of input examples, here three rows of two features each: [[a,b], [c,d], [e,f]].
A is incorrect: signature_name misspelled (seving_default) and malformed instances syntax.
B is incorrect: instances structure uses six separate scalars without proper grouping into rows, and still includes the typo in A.
C is incorrect: instances are grouped as two rows of three features each, not matching the expected two-feature pairs per instance.



Your organization's call center has asked you to develop a model that analyzes customer sentiments in each call. The call center receives over one million calls daily, and data is stored in Cloud Storage. The data collected must not leave the region in which the call originated, and no Personally Identifiable Information (PII) can be stored or analyzed. The data science team has a third-party tool for visualization and access which requires a SQL ANSI-2011 compliant interface. You need to select components for data processing and for analytics. How should the data pipeline be designed?

  1. 1= Dataflow, 2= BigQuery
  2. 1 = Pub/Sub, 2= Datastore
  3. 1 = Dataflow, 2 = Cloud SQL
  4. 1 = Cloud Function, 2= Cloud SQL

Answer(s): A

Explanation:

Option A is correct because:
- Dataflow (1) enables scalable, regionalized processing directly on data stored in Cloud Storage without moving data off its origin region, supporting compliant ETL/ML preprocessing for sentiment analysis.
- BigQuery (2) offers a ANSI-2011 compliant SQL interface via BI Engine and BigQuery SQL for analytics, while keeping data in GCS-origin region and avoiding PII storage/analysis within the pipeline; it also suits large-scale analytics.
Incorrect:
B) Pub/Sub is messaging, not full ETL/analysis; Datastore is NoSQL, not ANSI-2011 SQL-compatible analytics.
C) Cloud SQL is transactional DB not ideal for massive analytics/ad-hoc querying at scale; Dataflow alone doesn’t meet ANSI-2011 analytics requirement as implemented.
D) Cloud Function is event-driven compute, not scalable batch analytics; Cloud SQL limitation persists for large-scale SQL analytics.



You are an ML engineer at a global shoe store. You manage the ML models for the company's website. You are asked to build a model that will recommend new products to the user based on their purchase behavior and similarity with other users. What should you do?

  1. Build a classification model
  2. Build a knowledge-based filtering model
  3. Build a collaborative-based filtering model
  4. Build a regression model using the features as predictors

Answer(s): C

Explanation:

Option C is correct because collaborative filtering leverages user-item interactions to recommend products based on similarities between users or items, fitting purchase behavior and user likeness. A) Classification models predict discrete labels, not recommendations. B) Knowledge-based filtering relies on explicit domain rules and item attributes, not user interaction patterns. D) Regression models predict a continuous target, not ranking or recommending items based on user similarity.


Reference:

https://cloud.google.com/solutions/recommendations-using-machine-learning-on-compute-engine



  PDF
Viewing page 4 of 44
Viewing questions 25 - 32 out of 339 questions


Post your Comments and Discuss Google PROFESSIONAL MACHINE LEARNING ENGINEER exam prep with other Community members:

AI Tutor AI Tutor 👋 I’m here to help!