8 of the 40 questions in this set, with the correct answer marked and every option explained.
1. A machine learning team has several large CSV datasets in Amazon S3. Historically, models built with the Amazon SageMaker Linear Learner algorithm have taken hours to train on similar-sized datasets. The team’s leaders need to accelerate the training process. What can a machine learning specialist do to address this concern?
✓Use Amazon SageMaker Pipe mode.
Pipe mode streams training data directly from S3 to the algorithm container instead of first copying the full dataset onto the training instance's disk, which cuts the startup and I/O time that was making training slow.
✗Use AWS Glue to transform the CSV dataset to the JSON format.
Glue can reformat the files, but JSON is not a faster format for Linear Learner to ingest than CSV — the fix for slow training is changing how SageMaker reads the data (Pipe mode), not what format it's stored in.
✗Use Amazon Kinesis to stream the data to Amazon SageMaker.
Kinesis moves streaming data between producers and consumers in real time; the datasets here are static CSVs already sitting in S3, so there is no stream to build, and Kinesis does nothing to speed up how SageMaker reads training data.
✗Use Amazon Machine Learning to train the models.
Amazon Machine Learning was a simplified, now-retired AWS service for building predictive models from data already in S3, Redshift or RDS; it never addressed training speed and offered none of SageMaker's high-throughput data-loading modes.
2. A data engineer needs to set up a batch data ingestion pipeline that ingests data from an on-premises database into Amazon Redshift. Which service should the engineer use to accomplish this task?
✗Amazon Kinesis Data Streams
A low-level streaming service for continuous, real-time data producers and consumers — well suited to live event data, not a one-time batch load from an existing on-premises database.
✓AWS Database Migration Service (DMS)
DMS is purpose-built to replicate data from an external, on-premises database into an AWS target like Redshift, handling schema conversion and change capture along the way.
✗AWS Glue
A serverless ETL and data-cataloging service for batch and event-driven data transformation — it doesn't provide real-time low-latency stream compute on its own, train machine learning models, enforce per-row access control, or monitor a running job's health.
✗Amazon S3
Durable, scalable object storage — not a relational or NoSQL database, and on its own it has no compute engine to process real-time streams or run complex ETL transformations.
3. A data engineer needs to choose a storage solution for a highly dynamic dataset with frequent read and write operations. Which service is the most appropriate?
✗Amazon RDS
A managed relational database for structured, transactional (OLTP) workloads — not built for petabyte-scale analytics, key-value access at extreme scale, general object storage, time-series data, or scheduled infrastructure backups on its own.
✗Amazon Redshift
A petabyte-scale SQL data warehouse for analytical queries — not built for low-latency transactional access, general object storage, machine learning model training or deployment, or workflow/job orchestration on its own.
✗Amazon S3
Durable, scalable object storage — not a relational or NoSQL database, and on its own it has no compute engine to process real-time streams or run complex ETL transformations.
✓Amazon DynamoDB
DynamoDB is a managed NoSQL database designed for exactly this profile: very high request rates with single-digit-millisecond latency on both reads and writes.
4. A data engineer needs to automate the deployment of a data processing pipeline. Which AWS service is suitable for this task?
✗Amazon Redshift
A petabyte-scale SQL data warehouse for analytical queries — not built for low-latency transactional access, general object storage, machine learning model training or deployment, or workflow/job orchestration on its own.
✗Amazon Athena
A serverless SQL query engine for data already in S3 — built for ad hoc querying, not for training machine learning models, automating deployments, enforcing fine-grained access control, cataloging metadata, or processing real-time streams on its own.
✓AWS CodePipeline
CodePipeline automates the build-test-deploy stages of a pipeline's code and configuration, which is what 'automate the deployment' calls for.
✗Amazon EMR
A managed Hadoop/Spark cluster service for large-scale batch data processing — not a deployment-automation tool, a metadata catalog, a serverless query engine, or a lightweight real-time stream-analytics service.
5. Which AWS service can be used to manage encryption keys for data stored in AWS services?
✗AWS Shield
A DDoS-protection service for network- and application-layer attacks — a different security control from encryption-key management, account-activity threat detection, or sensitive-data discovery.
✓AWS Key Management Service (KMS)
KMS is AWS's dedicated service for creating, storing and controlling access to the cryptographic keys used to encrypt data across other AWS services.
✗AWS Secrets Manager
Stores and rotates secrets such as database credentials and API keys — a credential-management tool, not a data-masking, encryption-key-management, data-discovery, or web-application-firewall service.
✗Amazon GuardDuty
Continuously analyzes account activity for malicious or unauthorized behavior — a threat-detection service, not an encryption-key manager, sensitive-data classifier, web-application firewall, or configuration-compliance tracker.
6. A company wants to integrate data from various sources into a data lake on AWS. Which AWS service can be used to catalog, search, and query metadata from these sources?
✓AWS Glue Data Catalog
The Glue Data Catalog is AWS's central, persistent metadata store: Glue crawlers scan sources and populate it automatically, making the data catalogued, searchable and queryable from one place.
✗Amazon S3 Inventory
Reports on the objects and metadata already stored in a bucket — a bucket auditing tool, not a searchable catalog spanning multiple data sources.
✗Amazon Elasticsearch
A search and log-analytics engine — powerful for querying documents you've indexed into it, but it doesn't automatically crawl and catalog metadata across a data lake's sources the way Glue Data Catalog does.
✗AWS Data Pipeline
An older, more manual AWS service for scheduling data movement and transformation between stores — largely superseded by AWS Glue, which automates far more of the ETL, cataloging and orchestration work.
7. Which AWS service is primarily used to create and manage a secure data lake?
✗AWS Glue
A serverless ETL and data-cataloging service for batch and event-driven data transformation — it doesn't provide real-time low-latency stream compute on its own, train machine learning models, enforce per-row access control, or monitor a running job's health.
✗Amazon Redshift
A petabyte-scale SQL data warehouse for analytical queries — not built for low-latency transactional access, general object storage, machine learning model training or deployment, or workflow/job orchestration on its own.
✗Amazon EMR
A managed Hadoop/Spark cluster service for large-scale batch data processing — not a deployment-automation tool, a metadata catalog, a serverless query engine, or a lightweight real-time stream-analytics service.
✓AWS Lake Formation
Lake Formation is AWS's purpose-built service for setting up a data lake and enforcing fine-grained, centralized security and access control over it.
8. A company needs to process streaming data in real-time and store the processed data in Amazon S3. Which combination of services should the company use?
✗Amazon S3 and Amazon RDS
Neither is a streaming service — this pairing describes batch storage and transactional querying, with no mechanism to ingest continuous real-time data.
✗Amazon DynamoDB and AWS Data Pipeline
DynamoDB stores discrete items rather than an incoming stream, and Data Pipeline is a scheduled batch orchestrator, not a real-time processing engine.
✓Amazon Kinesis Data Streams and AWS Lambda
Kinesis Data Streams ingests the continuous events in real time, and Lambda processes each record as it arrives and writes the result to S3 — a fully managed pipeline with no servers to run.
✗Amazon Redshift and AWS Glue
Both are built around batch loads and scheduled jobs — neither ingests and processes continuous streaming data in real time.
32 more questions in the app
Practise the full 40-question set with a timer, scoring and progress tracking.
