Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Practice Test 38

100% Latest & Updated Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!

Question 370

A single ML platform team is handling two distinct changes for the same system. Workstream A: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. Workstream B: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. Select exactly TWO measures, one that directly resolves each workstream.

Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks.
Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.
Fix input throughput or select a more appropriate training instance before increasing GPU size.
Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.
Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity.

Explanation

Correct Answers

This action fits the situation: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. It implements the needed technical control: Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks. The intended outcome is that typed reusable code produces reviewable infrastructure deployments. The important limitation is that cDK synthesis does not eliminate the need to inspect the resulting cloud permissions or cost.

This action fits the situation: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. It implements the needed technical control: Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups. The intended outcome is that approved workloads can access the managed service over private network paths. The important limitation is that a private VPC endpoint for the API is not equivalent to attaching the endpoint's containers to arbitrary customer subnets.

Incorrect Answers

That alternative addresses a different task: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. The selected workloads instead require Typed infrastructure CDK, Private endpoint networking.

That alternative addresses a different task: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. The selected workloads instead require Typed infrastructure CDK, Private endpoint networking.

That alternative addresses a different task: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. The selected workloads instead require Typed infrastructure CDK, Private endpoint networking.

Question 371

A single ML platform team is handling two distinct changes for the same system. Workstream A: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. Workstream B: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. Select exactly TWO measures, one that directly resolves each workstream.

Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Fix input throughput or select a more appropriate training instance before increasing GPU size.
Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.
Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks.
Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.

Explanation

Correct Answers

This action fits the situation: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. It implements the needed technical control: Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity. The intended outcome is that serving capacity adjusts to sustained request load without permanently running peak instances. The important limitation is that scale-out reaction time can lag a sudden burst; minimum capacity still matters.

This action fits the situation: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. It implements the needed technical control: Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions. The intended outcome is that the private training environment can read approved S3 objects without relying on internet gateways. The important limitation is that a network endpoint policy cannot replace IAM and S3 bucket permission requirements.

Incorrect Answers

That alternative addresses a different task: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. The selected workloads instead require Invocation based endpoint scaling, Private S3 access for training.

That alternative addresses a different task: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. The selected workloads instead require Invocation based endpoint scaling, Private S3 access for training.

That alternative addresses a different task: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. The selected workloads instead require Invocation based endpoint scaling, Private S3 access for training.

Question 372

A single ML platform team is handling two distinct changes for the same system. Workstream A: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. Workstream B: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. Select exactly TWO measures, one that directly resolves each workstream.

Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.
Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity.
Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.
Enable managed Spot training and store durable checkpoints for resumption.

Explanation

Correct Answers

This action fits the situation: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. It implements the needed technical control: Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity. The intended outcome is that enough execution capacity can remain warm to serve time-sensitive bursts faster. The important limitation is that provisioned concurrency adds cost and does not imply an unlimited burst capacity.

This action fits the situation: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. It implements the needed technical control: Enable managed Spot training and store durable checkpoints for resumption. The intended outcome is that eligible training time can cost less while preserving recovery after interruptions. The important limitation is that spot capacity is not guaranteed and is unsuitable for hard completion deadlines without fallback.

Incorrect Answers

That alternative addresses a different task: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. The selected workloads instead require Serverless warm capacity, Spot training interruption strategy.

That alternative addresses a different task: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. The selected workloads instead require Serverless warm capacity, Spot training interruption strategy.

That alternative addresses a different task: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. The selected workloads instead require Serverless warm capacity, Spot training interruption strategy.

Question 373

A single ML platform team is handling two distinct changes for the same system. Workstream A: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. Workstream B: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. Select exactly TWO measures, one that directly resolves each workstream.

Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Fix input throughput or select a more appropriate training instance before increasing GPU size.
Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.
Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.
Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks.

Explanation

Correct Answers

This action fits the situation: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. It implements the needed technical control: Fix input throughput or select a more appropriate training instance before increasing GPU size. The intended outcome is that training capacity better matches the actual bottleneck and can reduce spend. The important limitation is that adding GPU accelerators does not fix slow host-side input decoding.

This action fits the situation: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. It implements the needed technical control: Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones. The intended outcome is that the service can continue accepting traffic when a host or zone has a localized problem. The important limitation is that one endpoint with only one instance cannot guarantee availability during a loss of that instance.

Incorrect Answers

That alternative addresses a different task: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. The selected workloads instead require Training compute right sizing, Resilient inference across zones.

That alternative addresses a different task: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. The selected workloads instead require Training compute right sizing, Resilient inference across zones.

That alternative addresses a different task: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. The selected workloads instead require Training compute right sizing, Resilient inference across zones.

Question 374

A cloud ML change request describes the following issue: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. Which action best addresses the actual requirement?

Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.
Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.
Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks.
Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. It does not solve the presented constraint: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. The matching action for this case is: Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.

This action is appropriate for another requirement: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. It does not solve the presented constraint: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. The matching action for this case is: Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.

This action is appropriate for another requirement: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. It does not solve the presented constraint: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. The matching action for this case is: Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.

Question 375

A team is choosing a specific control for this workload: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. Which action best addresses the actual requirement?

Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity.
Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.
Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.
Define versioned AWS CDK constructs and synthesize them into approved CloudFormation stacks.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. It does not solve the presented constraint: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. The matching action for this case is: Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.

This action is appropriate for another requirement: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. It does not solve the presented constraint: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. The matching action for this case is: Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.

This action is appropriate for another requirement: An application team wants reusable infrastructure constructs and automated checks using its existing TypeScript codebase rather than hand-editing long YAML definitions. It does not solve the presented constraint: A training job running within private subnets cannot download its input data because no public internet route or VPC endpoint is available. The matching action for this case is: Add an authorized S3 gateway endpoint or supported private access path and required bucket permissions.

Question 376

An ML engineering team faces this production constraint: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. Which action best addresses the actual requirement?

Enable managed Spot training and store durable checkpoints for resumption.
Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.
Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. It does not solve the presented constraint: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. The matching action for this case is: Enable managed Spot training and store durable checkpoints for resumption.

This action is appropriate for another requirement: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. It does not solve the presented constraint: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. The matching action for this case is: Enable managed Spot training and store durable checkpoints for resumption.

This action is appropriate for another requirement: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. It does not solve the presented constraint: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. The matching action for this case is: Enable managed Spot training and store durable checkpoints for resumption.

Question 377

A cloud ML change request describes the following issue: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. Which action best addresses the actual requirement?

Enable managed Spot training and store durable checkpoints for resumption.
Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.
Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. It does not solve the presented constraint: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. The matching action for this case is: Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.

This action is appropriate for another requirement: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. It does not solve the presented constraint: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. The matching action for this case is: Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.

This action is appropriate for another requirement: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. It does not solve the presented constraint: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. The matching action for this case is: Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.

Question 378

A team is choosing a specific control for this workload: Each Monday at 9 AM an online scoring service sees a predictable traffic jump that outpaces reactive auto scaling during the first minutes. Which action best addresses the actual requirement?

Add scheduled scaling actions to raise endpoint minimum desired capacity before the expected peak.
Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.
Fix input throughput or select a more appropriate training instance before increasing GPU size.
Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity.

Explanation

Correct Answer

This action fits the situation: Each Monday at 9 AM an online scoring service sees a predictable traffic jump that outpaces reactive auto scaling during the first minutes. It implements the needed technical control: Add scheduled scaling actions to raise endpoint minimum desired capacity before the expected peak. The intended outcome is that serving capacity is in place when the predictable workload begins. The important limitation is that scheduled capacity must be revised as usage patterns change, and overprovisioning raises cost.

Incorrect Answers

This action is appropriate for another requirement: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. It does not solve the presented constraint: Each Monday at 9 AM an online scoring service sees a predictable traffic jump that outpaces reactive auto scaling during the first minutes. The matching action for this case is: Add scheduled scaling actions to raise endpoint minimum desired capacity before the expected peak.

This action is appropriate for another requirement: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. It does not solve the presented constraint: Each Monday at 9 AM an online scoring service sees a predictable traffic jump that outpaces reactive auto scaling during the first minutes. The matching action for this case is: Add scheduled scaling actions to raise endpoint minimum desired capacity before the expected peak.

This action is appropriate for another requirement: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. It does not solve the presented constraint: Each Monday at 9 AM an online scoring service sees a predictable traffic jump that outpaces reactive auto scaling during the first minutes. The matching action for this case is: Add scheduled scaling actions to raise endpoint minimum desired capacity before the expected peak.

Question 379

An ML engineering team faces this production constraint: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. Which action best addresses the actual requirement?

Configure appropriate SageMaker serverless inference provisioned concurrency for predictable warm capacity.
Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.
Configure endpoint auto scaling on a tested invocations-per-instance target with suitable minimum and maximum capacity.
Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.

Explanation

Correct Answer

This action fits the situation: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. It implements the needed technical control: Publish approved CloudFormation outputs and import supported identifiers into the dependent stack. The intended outcome is that application infrastructure can reuse platform resources without copying unmanaged IDs. The important limitation is that cross-stack exports restrict some deletions or changes while another stack imports them.

Incorrect Answers

This action is appropriate for another requirement: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. It does not solve the presented constraint: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. The matching action for this case is: Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.

This action is appropriate for another requirement: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. It does not solve the presented constraint: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. The matching action for this case is: Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.

This action is appropriate for another requirement: An online model is overloaded during lunch-hour spikes while endpoint instances sit mostly idle overnight. It does not solve the presented constraint: A platform team manages networking in one AWS CloudFormation stack and SageMaker inference resources in another, and needs a controlled reference to an existing subnet ID. The matching action for this case is: Publish approved CloudFormation outputs and import supported identifiers into the dependent stack.

Question 380

During a production readiness audit, three independent ML pipeline behaviors must be verified. Workstream A: Three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. Workstream B: A regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. Workstream C: A customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. Select exactly THREE observations, one diagnostic that directly tests each workstream.

Inspect cold-start duration and supported provisioned concurrency settings for serverless inference.
Review GPU utilization, CPU preprocessing time, and input-pipeline throughput across representative batches.
Inspect endpoint healthy instance placement and traffic response during simulated instance failure.
Compare AWS CloudFormation stack changesets and resource drift across environments.
Inspect VPC interface endpoints, endpoint configuration, security groups, and DNS resolution.
Review interruptions, checkpoint availability, and managed Spot training savings relative to On-Demand.

Explanation

Correct Answers

This evidence directly tests the reported behavior: Inspect endpoint healthy instance placement and traffic response during simulated instance failure. For this case, a customer-facing model service requires continued availability after a single Availability Zone problem and relies on a managed SageMaker real-time endpoint. The results tell the team whether to apply the targeted change: Configure sufficient endpoint instance capacity and suitable managed hosting resilience across eligible availability zones.

This evidence directly tests the reported behavior: Compare AWS CloudFormation stack changesets and resource drift across environments. For this case, three environments require matching SageMaker endpoints, IAM roles, and logging resources managed through reviewed version-controlled templates. The results tell the team whether to apply the targeted change: Express the infrastructure in CloudFormation templates and deploy reviewed stack updates.

This evidence directly tests the reported behavior: Inspect VPC interface endpoints, endpoint configuration, security groups, and DNS resolution. For this case, a regulated SageMaker endpoint must be reachable only through authorized internal application traffic and cannot expose inference calls publicly. The results tell the team whether to apply the targeted change: Use supported private networking and VPC interface endpoints for SageMaker API or runtime access with restrictive security groups.

Incorrect Answers

That alternative addresses a different task: A low-traffic model has occasional latency-sensitive bursts, but its first invocation after idle time suffers from cold initialization. The selected workloads instead require CloudFormation reproducible endpoint, Private endpoint networking, Resilient inference across zones.

That alternative addresses a different task: A PyTorch training job uses a large GPU instance even though monitoring shows the GPU is idle while CPU data decoding is saturated. The selected workloads instead require CloudFormation reproducible endpoint, Private endpoint networking, Resilient inference across zones.

That alternative addresses a different task: A cost-constrained image training experiment can tolerate restart and should consume spare managed training capacity where available. The selected workloads instead require CloudFormation reproducible endpoint, Private endpoint networking, Resilient inference across zones.

Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Exam Dumps, Practice Test Questions

Question 370

Explanation

Correct Answers

Incorrect Answers

Question 371

Explanation

Correct Answers

Incorrect Answers

Question 372

Explanation

Correct Answers

Incorrect Answers

Question 373

Explanation

Correct Answers

Incorrect Answers

Question 374

Explanation

Correct Answer

Incorrect Answers

Question 375

Explanation

Correct Answer

Incorrect Answers

Question 376

Explanation

Correct Answer

Incorrect Answers

Question 377

Explanation

Correct Answer

Incorrect Answers

Question 378

Explanation

Correct Answer

Incorrect Answers

Question 379

Explanation

Correct Answer

Incorrect Answers

Question 380

Explanation

Correct Answers

Incorrect Answers