AWS Certified Machine Learning Engineer – Associate MLA-C01 Practice Questions

1060 exam-accurate questions with explanations

info

Free Sample Questions

Showing 10 of 1060 questions. Get full access to all questions, detailed explanations, and study materials.

Content Domain 1: Data Preparation for Machine Learning (ML)

An education platform is preparing an ML solution for ticket priority. An education platform stores 600 million events of curated tabular data in Amazon S3. Analysts commonly read a small subset of columns with Athena. Which format should the ML engineer choose?

A Write the curated dataset in Apache Parquet format. check_circle

B Write one JSON object per row without compaction.

C Store the analytical dataset as newline-delimited text only.

D Write the curated dataset as uncompressed CSV files.

lightbulb

Explanation

A is correct. Parquet is columnar and supports efficient compression and column pruning for analytical reads. The strongest distractor is D. CSV is portable, but it usually scans more bytes for column-oriented analytics.

Content Domain 1: Data Preparation for Machine Learning (ML)

A regulated workload at an industrial manufacturer has the following constraint. A data lake contains S3 objects partitioned by event date. The engineer needs a serverless SQL query to validate a sample before training. Which service should be used?

A Use Amazon OpenSearch Service as the default SQL engine for S3.

B Use Amazon Athena to query the data in Amazon S3. check_circle

C Use Amazon Redshift provisioned clusters for a one-time inspection.

D Use AWS Glue DataBrew only to run SQL queries.

lightbulb

Explanation

B is correct. Athena provides serverless SQL querying directly over S3 data. The strongest distractor is C. A provisioned Redshift cluster can work for warehouse workloads, but it adds unnecessary management for a serverless S3 inspection.

Content Domain 1: Data Preparation for Machine Learning (ML)

An engineer is troubleshooting an AWS ML workload for an energy company. The ML team must catalog raw S3 data and run repeatable Spark-based transformations. Which AWS service should it select?

A Use Amazon Athena as the primary distributed ETL runtime.

B Use AWS Lake Formation as the Spark execution engine.

C Use AWS Glue and the AWS Glue Data Catalog. check_circle

D Use Amazon Data Firehose as the metadata catalog.

lightbulb

Explanation

C is correct. AWS Glue provides managed ETL and integrates with the Glue Data Catalog. The strongest distractor is A. Athena is useful for SQL queries, but it is not the primary distributed ETL runtime for this requirement.

Content Domain 1: Data Preparation for Machine Learning (ML)

The next release at a property-management platform includes a change to the ML workflow. Several training instances require concurrent POSIX-style access to shared feature files. Which storage category is appropriate?

A Use Amazon DynamoDB to emulate POSIX file locking.

B Use a single Amazon EBS volume attached to one instance.

C Use Amazon S3 Glacier Deep Archive as the active training mount.

D Use a shared file system such as Amazon EFS or Amazon FSx that matches the throughput requirements. check_circle

lightbulb

Explanation

D is correct. EFS or FSx can provide shared file-system access for training workloads. The strongest distractor is B. A single EBS volume is not the normal choice for concurrent shared file access across multiple training instances.

Content Domain 1: Data Preparation for Machine Learning (ML)

The platform team at a sports analytics company is improving a workflow for customer clickstream events. The ML platform needs windowed aggregations over an event stream before persisting features. Which managed service should it choose?

A Use Amazon Managed Service for Apache Flink. check_circle

B Use AWS Glue crawlers on a one-minute schedule.

C Use AWS DataSync to calculate streaming windows.

D Use SageMaker batch transform for each event.

lightbulb

Explanation

A is correct. Managed Service for Apache Flink is designed for stateful streaming transformations and windowed aggregations. The strongest distractor is B. Glue crawlers discover metadata; they do not perform stateful event-stream processing.

Content Domain 1: Data Preparation for Machine Learning (ML)

An engineer is troubleshooting an AWS ML workload for a cybersecurity vendor. The team wants a managed stream-delivery path into S3 and does not need complex stateful processing. What is the best fit?

A Use Amazon EBS snapshots as the delivery mechanism.

B Use Amazon Data Firehose to deliver the stream to Amazon S3. check_circle

C Use Amazon Managed Service for Apache Flink for every simple delivery use case.

D Use AWS DataSync for event-by-event stream ingestion.

lightbulb

Explanation

B is correct. Amazon Data Firehose is suited to managed stream delivery to destinations such as S3 with minimal operational effort. The strongest distractor is C. Flink is powerful for stateful processing, but it is unnecessarily complex when the main need is managed delivery.

Content Domain 1: Data Preparation for Machine Learning (ML)

During a design review at a telecom operator, the team identifies the following requirement. A low-cardinality feature contains non-ordinal values such as product category. Which encoding is appropriate for a linear model?

A Drop the feature automatically because it is categorical.

B Apply label encoding and treat the integer values as ordered magnitudes.

C Apply one-hot encoding. check_circle

D Replace each category with a random hash and assume the hash magnitude is meaningful.

lightbulb

Explanation

C is correct. One-hot encoding represents low-cardinality non-ordinal categories without introducing a false ranking. The strongest distractor is B. Direct label encoding can incorrectly imply an ordinal relationship.

Content Domain 1: Data Preparation for Machine Learning (ML)

The platform team at a video-streaming platform is improving a workflow for marketing-response events. A data scientist wants to profile a dataset and create reusable preprocessing steps visually inside a SageMaker-oriented workflow. What is the best fit?

A Use Amazon Athena only as the visual transformation interface.

B Use AWS Config rules to impute missing values.

C Use SageMaker Model Monitor to clean the training data interactively.

D Use Amazon SageMaker Data Wrangler. check_circle

lightbulb

Explanation

D is correct. SageMaker Data Wrangler supports exploration, profiling, and data transformation for ML preparation. The strongest distractor is C. Model Monitor evaluates production data and model behavior; it is not the interactive preparation tool for this requirement.

Content Domain 1: Data Preparation for Machine Learning (ML)

An engineer is troubleshooting an AWS ML workload for a video-streaming platform. A data analyst needs a visual, no-code-oriented tool to clean tabular data before handing it to the ML team. Which service is appropriate?

A Use AWS Glue DataBrew. check_circle

B Use AWS Glue Data Quality only as the visual cleaning tool.

C Use Amazon QuickSight to alter source records.

D Use AWS Lake Formation to normalize feature values.

lightbulb

Explanation

A is correct. Glue DataBrew provides visual data preparation and cleaning recipes. The strongest distractor is B. Glue Data Quality validates data rules, but it is not the primary visual cleaning interface.

Content Domain 1: Data Preparation for Machine Learning (ML)

A cost and reliability review at a property-management platform raises the following question. A team wants to reduce training-serving skew by reusing governed features in batch training and low-latency inference. What should it use?

A Use AWS Secrets Manager as the offline feature repository.

B Use SageMaker Feature Store with the appropriate online and offline stores. check_circle

C Use Amazon Athena query history as the online feature store.

D Store features only in notebook-local CSV files.

lightbulb

Explanation

B is correct. SageMaker Feature Store supports managed feature reuse for online and offline access patterns. The strongest distractor is D. Notebook-local files do not provide a governed shared feature repository or production online access.

Get all 1060 questions

Full access includes all questions, detailed explanations, PDF downloads, and timed mock exams.

quiz Start Free Practice download Free PDF