A company is building an Amazon SageMaker AI pipeline for an ML model. The pipeline uses distributed processing and training.
An ML engineer needs to encrypt network communication between instances that run distributed jobs. The ML engineer configures the distributed jobs to run in a private VPC.
What should the ML engineer do to meet the encryption requirement?
A. Enable network isolation.
B. Configure traffic encryption by using security groups.
C. Enable inter-container traffic encryption.
D. Enable VPC flow logs.
Show Answer
Correct Answer: C
Explanation: For Amazon SageMaker distributed training and processing jobs, inter-container traffic encryption encrypts network communication between the instances/containers participating in the distributed job. Running the jobs in a private VPC does not by itself encrypt traffic. Network isolation restricts container network access, security groups filter traffic but do not encrypt it, and VPC Flow Logs only record network metadata.
Question 82
HOTSPOT
-
An airline company deploys ML models to one dozen Amazon SageMaker AI inference endpoints. The inference endpoints must be able to handle different types of workloads in a cost-effective way.
Select the correct inference option from the following list to handle each type of workload. Select each inference option one time. (Choose four.)
• Asynchronous inference
• Batch inference
• Real-time inference
• Serverless inference
Explanation: Use real-time for low-latency responses, serverless for spiky unpredictable traffic, batch for large offline datasets, and asynchronous for long-running inference such as image/audio generation.
Question 83
A company needs to analyze a large dataset that is stored in Amazon S3 in Apache Parquet format. The company wants to use one-hot encoding for some of the columns.
The company needs a no-code solution to transform the data. The solution must store the transformed data back to the same S3 bucket for model training.
Which solution will meet these requirements?
A. Configure an AWS Glue DataBrew project that connects to the data. Use the DataBrew interactive interface to create a recipe that performs the one-hot encoding transformation. Create a job to apply the transformation and to write the output back to an S3 bucket.
B. Configure an AWS Glue Data Catalog table that points to the data. Use Amazon Athena to write SQL commands to perform the one-hot encoding transformation. Configure Athena to write the query results back to an S3 bucket.
C. Configure an AWS Glue Data Catalog table that points to the data. Create an AWS Glue ETL interactive notebook. Use the notebook to perform the one-hot encoding transformation. Run the configured cells and write the results back to an S3 bucket.
D. Configure an Amazon Redshift cluster to access the data by using Redshift Spectrum. Use SQL commands to perform the one-hot encoding transformation within Amazon Redshift. Configure Amazon Redshift to write the results back to an S3 bucket.
Show Answer
Correct Answer: A
Explanation: AWS Glue DataBrew is a no-code data preparation service that provides built-in transformations such as one-hot encoding through a visual interface. It can read Parquet data from Amazon S3, apply a recipe as a job, and write the transformed output back to Amazon S3. The other options require coding or SQL and therefore do not satisfy the no-code requirement.
Question 84
An ecommerce company trains an ML model to forecast demand for near real-time inventory management based on historical customer activity. The company successfully deploys the trained model to a production Amazon SageMaker AI endpoint. However, the company notices that the model’s forecast performance degrades over time. The company needs a long-term and automated solution to mitigate the performance degradation.
Which solution will meet these requirements?
A. Use Amazon SageMaker Debugger to automatically send alerts when model performance anomalies are detected.
B. Use AWS X-Ray to monitor the performance of the SageMaker AI endpoint and the incoming requests to inform model re-training.
C. Use Amazon SageMaker Ground Truth to curate a high-quality dataset. Use the dataset to re-train the model.
D. Use Amazon SageMaker Clarify to monitor model and feature attribution bias to inform model re-training.
Show Answer
Correct Answer: D
Explanation: The issue is model performance degrading over time due to changing production data. SageMaker Clarify can monitor data drift and feature attribution drift in production, providing ongoing detection of distribution changes that inform automated retraining workflows. SageMaker Debugger focuses on debugging and profiling training and inference rather than monitoring data/model drift for long-term model quality. Ground Truth is for data labeling, and X-Ray is for application tracing, not ML model performance.
Question 85
An ML engineer wants to use Amazon SageMaker AI to prepare data for training. During exploratory data analysis, the ML engineer notices that several categorical features are missing values.
How can the ML engineer use SageMaker AI to solve this problem?
A. Use SageMaker Clarify to impute categorical features with the mean value.
B. Use SageMaker Clarity to impute categorical features with the mode value.
C. Use SageMaker Data Wrangler to impute categorical features with the mean value.
D. Use SageMaker Data Wrangler to impute categorical features with the mode value.
Show Answer
Correct Answer: D
Explanation: Amazon SageMaker Data Wrangler is the SageMaker service designed for data preparation and feature engineering, including handling missing values through imputation. For categorical features, the appropriate imputation strategy is the mode (most frequent value), whereas the mean is only appropriate for numerical features. SageMaker Clarify is used for bias detection and model explainability, not data imputation.
Question 86
A company has an ML model in Amazon SageMaker AI. An ML engineer needs to implement a monitoring solution to automatically detect changes in the input data distribution of model features.
Which solution will meet this requirement with the LEAST operational overhead?
A. Configure SageMaker Model Monitor. Establish a data quality baseline. Ensure that the emit_metrics option is enabled in the baseline constraints file. Configure an Amazon CloudWatch alarm to notify the company about changes in specific metrics that are related to data quality.
B. Configure SageMaker Model Monitor. Establish a model quality baseline. Ensure that the comparison_method option is set to Robust in the baseline constraints file. Configure an Amazon CloudWatch alarm to notify the company about changes in model quality metrics.
C. Use SageMaker Debugger with custom rules to track shifts in feature distributions. Configure Amazon CloudWatch alarms to notify the company when the rules detect significant changes.
D. Use Amazon CloudWatch to directly observe the SageMaker AI endpoint’s performance metrics. Manually analyze the CloudWatch logs for indicators of data drift or shifts in feature distribution.
Show Answer
Correct Answer: A
Explanation: SageMaker Model Monitor's data quality monitoring is designed to detect changes in input feature distributions by comparing live inference data against a data quality baseline. Enabling metric emission allows CloudWatch metrics and alarms for automated notification with minimal operational overhead. Model quality monitoring tracks prediction performance rather than input drift, Debugger is not the managed solution for production data drift monitoring, and manual CloudWatch log analysis has higher operational overhead.
Question 87
A company collects customer data every day. The company stores the data as compressed files in an Amazon S3 bucket that is partitioned by date. Every month, analysts download the data, process the data to check the data quality, and then upload the data to Amazon QuickSight dashboards.
An ML engineer needs to implement a solution to automatically check the data quality before the data is sent to QuickSight.
Which solution will meet these requirements with the LEAST operational overhead?
A. Run an AWS Glue crawler every month to update the AWS Glue Data Catalog. Use AWS Glue Data Quality rules to check the data quality.
B. Use an AWS Glue trigger to run an AWS Glue crawler every month to update the AWS Glue Data Catalog. Create an AWS Glue job that loads the data into a PySpark DataFrame. Configure the job to apply custom functions and to evaluate the data quality.
C. Run Python scripts on an AWS Lambda function every month to evaluate data quality. Configure the S3 bucket to invoke the Lambda function when objects are added to the S3 bucket.
D. Configure the S3 bucket to send event notifications to an Amazon Simple Queue Service (Amazon SQS) queue when objects are uploaded. Use Amazon CloudWatch insights every month for the SQS queue to evaluate the data quality.
Show Answer
Correct Answer: A
Explanation: AWS Glue Data Quality is a fully managed feature designed to profile and validate datasets using declarative data quality rules. Running a Glue crawler to keep the Data Catalog current and then using Glue Data Quality minimizes custom code and operational overhead. Option B requires maintaining custom PySpark jobs, option C relies on custom Lambda scripts and is not aligned with the monthly processing workflow, and option D is incorrect because CloudWatch Logs Insights does not perform data quality evaluation for SQS messages.
Question 88
An ML model is deployed in production. The model has performed well and has met its metric thresholds for months.
An ML engineer who is monitoring the model observes a sudden degradation. The performance metrics of the model are now below the thresholds.
What could be the cause of the performance degradation?
A. Lack of training data
B. Drift in production data distribution
C. Compute resource constraints
D. Model overfitting
Show Answer
Correct Answer: B
Explanation: A model that has performed well for months and then suddenly degrades is most commonly experiencing data drift (or concept drift), where the production data distribution has changed from the training distribution. Lack of training data and overfitting are training-time issues that would typically appear earlier rather than after months of stable production. Compute resource constraints may affect latency or availability but do not usually cause a sustained drop in model accuracy metrics.
Question 89
A hospital is using an ML model to validate x-ray results. The hospital runs a nightly batch inference job. The hospital needs to produce a daily report about model data quality and model performance.
Which solution will meet these requirements?
A. Schedule a monitoring job in Amazon SageMaker Model Monitor. Generate the monitoring results for the model and data.
B. Create an Amazon CloudWatch dashboard that includes the metrics for processing steps in the nightly batch inference job. Compare the baseline resource metrics. Share the dashboard link.
C. Use AWS Glue DataBrew to create a custom recipe job that uses the Numerical Statistics data quality check for the model file. Generate the results.
D. Create a SageMaker AI pipeline that includes a QualityCheck step to run monitoring jobs. Generate the monitoring results for the model and the data.
Show Answer
Correct Answer: D
Explanation: A SageMaker AI Pipeline with a QualityCheck step can run model/data quality monitoring as part of the nightly batch workflow and produce monitoring reports after each batch inference run. This is well suited for scheduled batch inference pipelines that need daily reports on both data quality and model performance. A scheduled Model Monitor job alone is primarily for ongoing monitoring and does not by itself orchestrate the end-to-end batch workflow with quality check steps.
Question 90
A company runs an Amazon SageMaker AI domain in a public subnet of a newly created VPC. The network is configured properly, and ML engineers can access the SageMaker AI domain.
Recently, the company discovered suspicious traffic to the domain from a specific IP address. The company needs to block traffic from the specific IP address.
Which update to the network configuration will meet this requirement?
A. Create a security group inbound rule to deny traffic from the specific IP address. Assign the security group to the domain.
B. Create a network ACL inbound rule to deny traffic from the specific IP address. Assign the rule to the default network ACL for the subnet where the domain is located.
C. Create a shadow variant for the domain. Configure SageMaker Inference Recommender to send traffic from the specific IP address to the shadow endpoint.
D. Create a VPC route table to deny inbound traffic from the specific IP address. Assign the route table to the domain.
Show Answer
Correct Answer: B
Explanation: Network ACLs support explicit allow and deny rules at the subnet level, including blocking a specific source IP address. Security groups are stateful and do not support deny rules. Route tables do not filter traffic by source IP, and SageMaker shadow variants are unrelated to network access control.
$19
Get all 235 questions with detailed answers and explanations
Instant download HTML + PDF delivered the moment payment clears.
Secure Stripe checkout we never see or store your card details.
7-day refund if files are defective see our refund policy.
