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
CrowdStrike
Anthropic
All Vendors
  1. Home
  2. Exams
  3. Google
  4. PROFESSIONAL MACHINE LEARNING ENGINEER

Free PROFESSIONAL MACHINE LEARNING ENGINEER Exam Braindumps
Professional Machine Learning Engineer Questions 101–105 (page: 22)

Updated On: 25-Oct-2026
Page 22 of 82
PDF with 402 Questions

You manage a team of data scientists who use a cloud-based backend system to submit training jobs. This system has become very difficult to administer, and you want to use a managed service instead. The data scientists you work with use many different frameworks, including Keras, PyTorch, theano, scikit-learn, and custom libraries.
What should you do?

  1. Use the Vertex AI Training to submit training jobs using any framework.
  2. Configure Kubeflow to run on Google Kubernetes Engine and submit training jobs through TFJob.
  3. Create a library of VM images on Compute Engine, and publish these images on a centralized repository.
  4. Set up Slurm workload manager to receive jobs that can be scheduled to run on your cloud infrastructure.

Answer(s): A



You manage an ML workflow for a fraud detection model. The source code for data processing and model training is stored in a version control repository. You need to implement a continuous training (CT) pipeline that automatically runs unit tests and launches a new training run on Agent Platform Pipelines when code is updated in the repository. You want to use a fully managed service to orchestrate this process.
What should you do?

  1. Create a Cloud Run function that is triggered by repository commits. The function runs the unit tests, launches a custom training job, and uses Model Monitoring to validate the new model’s performance before deployment.
  2. Define a Cloud Workflow to monitor the repository for changes. Configure the workflow on Agent Platform Pipelines to orchestrate and run the unit tests. Use the model evaluation component to validate the new model’s performance before deployment.
  3. Configure Agent Platform Pipelines to poll the repository for updates. Add a conditional step in the pipeline that uses Model Monitoring to validate the new model’s performance before deployment.
  4. Create a Cloud Build trigger associated with the repository. Define a build configuration file that runs the unit tests and triggers a job on Agent Platform Pipelines. Use the model evaluation component to validate the model performance before deployment.

Answer(s): D

Explanation:

Cloud Build provides fully managed, repository-triggered CI automation. A build trigger can run unit tests whenever code changes are committed and, after successful validation, invoke the Agent Platform Pipeline for retraining. The pipeline can then evaluate the new model before allowing deployment.



Your team is training a large number of ML models that use different algorithms, parameters, and datasets. Some models are trained in Vertex AI Pipelines, and some are trained on Vertex AI Workbench notebook instances. Your team wants to compare the performance of the models across both services. You want to minimize the effort required to store the parameters and metrics.
What should you do?

  1. Implement an additional step for all the models running in pipelines and notebooks to export parameters and metrics to BigQuery.
  2. Create a Vertex AI experiment. Submit all the pipelines as experiment runs. For models trained on notebooks log parameters and metrics by using the Vertex AI SDK.
  3. Implement all models in Vertex AI Pipelines Create a Vertex AI experiment, and associate all pipeline runs with that experiment.
  4. Store all model parameters and metrics as model metadata by using the Vertex AI Metadata API.

Answer(s): B



You are orchestrating a retraining pipeline for a custom tabular model using Agent Platform Pipelines. You need to ensure that the newly ingested data matches the expected schema and that its feature distribution has not drifted significantly from the baseline dataset used in the previous model version. You need to ensure that the training component executes only if these conditions are met.
What should you do?

  1. Ingest the new data into Agent Platform Feature Store using a batch import job. Configure Agent Platform Feature Store to automatically reject the import operation if the input data schema does not match the existing entity type definitions.
  2. Add a pipeline component that generates statistics for the new dataset and compares them against the baseline dataset's statistics. Use a conditional step to proceed to the training component only if no anomalies are detected.
  3. Configure a Model Monitoring job to analyze the training dataset for feature skew and data drift relative to the previous training run. Set an alert threshold that triggers a Cloud Run function to stop new Agent Platform Pipelines jobs from executing if drift is detected.
  4. Update the training application code to calculate feature distributions using a Python library. Add an assertion in the code after training to raise an exception and exit the application if the distributions differ from the baseline by a defined threshold.

Answer(s): B

Explanation:

Agent Platform Pipelines supports automated data validation and conditional execution to ensure that only high-quality, consistent data is used for model retraining. A pipeline component can generate statistical summaries of the newly ingested dataset and compare them with baseline statistics from the previous training dataset. This comparison identifies schema mismatches, missing values, and significant changes in feature distributions. Using a conditional pipeline step ensures that the training component executes only when the validation checks pass and no significant anomalies are detected. This approach prevents training on unsuitable data, reduces the risk of model performance degradation, and maintains a reliable, automated retraining workflow.



You are designing an Agent Platform Pipelines workflow for a real-time credit card fraud detection model. The model relies on dynamic features that change frequently, such as "number of transactions in the last hour," which are updated continuously in Agent Platform Feature Store. You are preparing the training dataset using historical transaction logs. You need to ensure that the feature values associated with each historical transaction in the training set accurately reflect the user's state at the exact moment the transaction occurred. You want to use the simplest approach.
What should you do?

  1. Configure the pipeline to run a batch serving job from Agent Platform Feature Store. Point the job to the transaction IDs and their timestamps to perform a point-in-time lookup to create the training dataset.
  2. Configure the pipeline to export the current feature values from Agent Platform Feature Store to BigQuery. Write a SQL query to join these values with the historical transaction logs to create the training dataset.
  3. Create a Dataflow pipeline that ingests the historical transaction logs, groups them by user, and calculates the latest aggregated statistics. Store the results in Cloud Storage for the training pipeline to consume.
  4. Create a Dataflow pipeline that reads the historical transaction logs. For each transaction, configure the pipeline to query the Agent Platform Feature Store online serving to retrieve the feature values. Join the feature values with the transaction data to create the training dataset.

Answer(s): A

Explanation:

Agent Platform Feature Store batch serving supports point-in-time feature retrieval, allowing the training pipeline to retrieve historical feature values associated with specific entity IDs and transaction timestamps. This ensures that each transaction is matched with the feature values available at the time it occurred, rather than their current values. Using point-in-time lookups prevents data leakage, preserves temporal consistency between training and inference, and improves fraud detection model reliability. Batch serving also provides a scalable, managed approach to generating historically accurate training datasets without requiring custom data processing pipelines.


Reference:

https://docs.cloud.google.cn/gemini-enterprise-agent-platform/reference/rest/v1beta1/projects.locat ions.featurestores/batchReadFeatureValues https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/main/notebooks/official/ feature_store_legacy/sdk-feature-store.ipynb?hl=he



  PDF
Viewing page 22 of 82
Viewing questions 106 - 110 out of 402 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!