Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Practice Test 111
Question 1100
During a production MLOps reliability and security review, the following condition is observed: A release pipeline downloads an ML model artifact from a mutable object key and deploys it without checking whether its bytes changed since approval. The owner needs an auditable resolution without broadening privileges. Which verification result BEST demonstrates that the issue is resolved?
- The deployed model bytes match the signed or approved manifest and a tampered artifact fails validation.
- A test write to an approved prefix succeeds, while writes to unapproved external destinations are denied.
- Peak-window 5xx rate and p95 latency fall below service objectives with verified alarms.
- Parity checks match transformations within documented freshness tolerances for test entities.
Explanation
Correct Answer
A:The deployed model bytes match the signed or approved manifest and a tampered artifact fails validation. This is the acceptance evidence for the described model artifact supply-chain integrity scenario. Object names alone do not guarantee model provenance; supply-chain checks bind deployment to reviewed bytes. The corrective implementation it validates is: Pin an immutable approved artifact version or content digest and verify integrity before deployment. Reported signal: A release pipeline downloads an ML model artifact from a mutable object key and deploys it without checking whether its bytes changed since approval. Operational setting: During a production MLOps reliability and security review.
Incorrect Answers
B:A test write to an approved prefix succeeds, while writes to unapproved external destinations are denied. This could validate cross-account s3 bucket exfiltration protection, but it does not establish that model artifact supply-chain integrity is resolved for this workload. Confirm instead: The deployed model bytes match the signed or approved manifest and a tampered artifact fails validation. Observed problem: A release pipeline downloads an ML model artifact from a mutable object key and deploys it without checking whether its bytes changed since approval. Operational setting: During a production MLOps reliability and security review.
C:Peak-window 5xx rate and p95 latency fall below service objectives with verified alarms. This could validate cloudwatch 5xxerrors endpoint saturation, but it does not establish that model artifact supply-chain integrity is resolved for this workload. Confirm instead: The deployed model bytes match the signed or approved manifest and a tampered artifact fails validation. Observed problem: A release pipeline downloads an ML model artifact from a mutable object key and deploys it without checking whether its bytes changed since approval. Operational setting: During a production MLOps reliability and security review.
D:Parity checks match transformations within documented freshness tolerances for test entities. This could validate training-serving feature skew incident, but it does not establish that model artifact supply-chain integrity is resolved for this workload. Confirm instead: The deployed model bytes match the signed or approved manifest and a tampered artifact fails validation. Observed problem: A release pipeline downloads an ML model artifact from a mutable object key and deploys it without checking whether its bytes changed since approval. Operational setting: During a production MLOps reliability and security review.
Question 1101
As a regional data-pipeline rollout is validated against approved AWS storage and streams, the following condition is observed: A new marketing campaign adds hundreds of thousands of unique user IDs, making naïve one-hot features extremely sparse and too large to maintain. The data team must preserve event-time meaning and lineage. Which investigation provides the most decisive FIRST evidence for the underlying problem?
- Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers.
- Inspect workload read/write semantics, shared access requirements, filesystem operations, and object-store request patterns.
- Inspect preprocessing fit order, imputation statistics, dataset split boundaries, and pipeline serialization.
- Inspect event identifiers, object versions or ETags, processing ledger, and downstream record keys for duplicate deliveries.
Explanation
Correct Answer
A:Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers. This directly tests high-cardinality categorical encoding in the reported AWS ML workflow. One-hot encoding every nearly unique identifier can waste capacity and overfit identities instead of learning transferable patterns. It distinguishes the problem before choosing a change. The related targeted response is: Drop leakage-prone IDs or adopt controlled hashing or suitable learned encodings with stable unknown-category handling. Reported signal: A new marketing campaign adds hundreds of thousands of unique user IDs, making naïve one-hot features extremely sparse and too large to maintain. Operational setting: As a regional data-pipeline rollout is validated against approved AWS storage and streams.
Incorrect Answers
B:Inspect workload read/write semantics, shared access requirements, filesystem operations, and object-store request patterns. This is a possible diagnostic for s3 versus efs training storage choice, but it does not directly test the high-cardinality categorical encoding failure reported here. Prioritize: Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers. The relevant case is: A new marketing campaign adds hundreds of thousands of unique user IDs, making naïve one-hot features extremely sparse and too large to maintain. Operational setting: As a regional data-pipeline rollout is validated against approved AWS storage and streams.
C:Inspect preprocessing fit order, imputation statistics, dataset split boundaries, and pipeline serialization. This is a possible diagnostic for median imputation fitted on training split, but it does not directly test the high-cardinality categorical encoding failure reported here. Prioritize: Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers. The relevant case is: A new marketing campaign adds hundreds of thousands of unique user IDs, making naïve one-hot features extremely sparse and too large to maintain. Operational setting: As a regional data-pipeline rollout is validated against approved AWS storage and streams.
D:Inspect event identifiers, object versions or ETags, processing ledger, and downstream record keys for duplicate deliveries. This is a possible diagnostic for s3 object event at-least-once processing, but it does not directly test the high-cardinality categorical encoding failure reported here. Prioritize: Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers. The relevant case is: A new marketing campaign adds hundreds of thousands of unique user IDs, making naïve one-hot features extremely sparse and too large to maintain. Operational setting: As a regional data-pipeline rollout is validated against approved AWS storage and streams.
Question 1102
As a regulated ML workload prepares for an automated regional rollout, the following condition is observed: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. The team must demonstrate a repeatable monitoring or security control. Which TWO steps best establish the evidence and correct the identified root cause? Select TWO.
- Use approved SageMaker interface endpoints and relevant private DNS and policy settings for required API access.
- Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types.
- Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting.
- Inspect group-level prediction and label distributions, subgroup sample sizes, and configured bias metrics.
- Scale compatible asynchronous inference capacity or tune workload and queue alarms for pending-request age.
Explanation
Correct Answers
B:Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types. This directly tests tag-based ml training cost attribution in the reported AWS ML workflow. Cost Explorer grouping requires supported activated resource tags; naming resources does not automatically allocate charges. It distinguishes the problem before choosing a change. The related targeted response is: Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting. This is one part of a select-two decision; the matching evidence and corrective action are both required. Reported signal: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. Operational setting: As a regulated ML workload prepares for an automated regional rollout.
C:Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting. This is the appropriate controlled remedy for tag-based ml training cost attribution. Cost Explorer grouping requires supported activated resource tags; naming resources does not automatically allocate charges. Verify the cause with: Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types. Then confirm: Cost reports attribute the majority of in-scope ML spend to the correct team without altering access controls. This is one part of a select-two decision; the matching evidence and corrective action are both required. Reported signal: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. Operational setting: As a regulated ML workload prepares for an automated regional rollout.
Incorrect Answers
A:Use approved SageMaker interface endpoints and relevant private DNS and policy settings for required API access. This selection concerns private vpc endpoint policy for sagemaker apis rather than the required diagnostic-and-remedy pair for tag-based ml training cost attribution. Use evidence: Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types. and the matching correction: Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting. Signal: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. Operational setting: As a regulated ML workload prepares for an automated regional rollout.
D:Inspect group-level prediction and label distributions, subgroup sample sizes, and configured bias metrics. This selection concerns clarify postdeployment group bias monitoring rather than the required diagnostic-and-remedy pair for tag-based ml training cost attribution. Use evidence: Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types. and the matching correction: Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting. Signal: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. Operational setting: As a regulated ML workload prepares for an automated regional rollout.
E:Scale compatible asynchronous inference capacity or tune workload and queue alarms for pending-request age. This selection concerns asynchronous inference queue backlog alert rather than the required diagnostic-and-remedy pair for tag-based ml training cost attribution. Use evidence: Inspect resource tags, cost allocation tag activation, Cost Explorer grouping, and unallocated resource types. and the matching correction: Apply consistent approved cost allocation tags to ML jobs and supported linked resources and activate relevant reporting. Signal: An organization cannot allocate expensive SageMaker training jobs among departments because jobs and dependent storage lack common cost allocation tags. Operational setting: As a regulated ML workload prepares for an automated regional rollout.
Question 1103
During production evaluation of a SageMaker AI model candidate, the following condition is observed: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. The release is blocked until the underlying model behavior is understood. Select TWO: the appropriate diagnostic check and the targeted implementation change.
- Deploy a SageMaker shadow variant for observational comparison before promoting the new model to served responses.
- Use an appropriate text classification model or supported SageMaker BlazingText supervised approach with validated labels.
- Inspect squared versus absolute residual contributions, target distribution, and large-error business costs.
- Inspect training job InputDataConfig, environment variables such as SM_CHANNEL_TRAINING, and container paths.
- Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses.
Explanation
Correct Answers
C:Inspect squared versus absolute residual contributions, target distribution, and large-error business costs. This directly tests rmse versus mae outlier sensitivity in the reported AWS ML workflow. Squaring residuals gives large errors disproportionately high weight; MAE weights magnitudes linearly. It distinguishes the problem before choosing a change. The related targeted response is: Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses. This is one part of a select-two decision; the matching evidence and corrective action are both required. Reported signal: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. Operational setting: During production evaluation of a SageMaker AI model candidate.
E:Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses. This is the appropriate controlled remedy for rmse versus mae outlier sensitivity. Squaring residuals gives large errors disproportionately high weight; MAE weights magnitudes linearly. Verify the cause with: Inspect squared versus absolute residual contributions, target distribution, and large-error business costs. Then confirm: The evaluation report shows both typical error and extreme-error impact with a justified primary metric. This is one part of a select-two decision; the matching evidence and corrective action are both required. Reported signal: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. Operational setting: During production evaluation of a SageMaker AI model candidate.
Incorrect Answers
A:Deploy a SageMaker shadow variant for observational comparison before promoting the new model to served responses. This selection concerns shadow deployment model evaluation without response change rather than the required diagnostic-and-remedy pair for rmse versus mae outlier sensitivity. Use evidence: Inspect squared versus absolute residual contributions, target distribution, and large-error business costs. and the matching correction: Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses. Signal: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. Operational setting: During production evaluation of a SageMaker AI model candidate.
B:Use an appropriate text classification model or supported SageMaker BlazingText supervised approach with validated labels. This selection concerns sagemaker built-in blazingtext choice rather than the required diagnostic-and-remedy pair for rmse versus mae outlier sensitivity. Use evidence: Inspect squared versus absolute residual contributions, target distribution, and large-error business costs. and the matching correction: Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses. Signal: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. Operational setting: During production evaluation of a SageMaker AI model candidate.
D:Inspect training job InputDataConfig, environment variables such as SM_CHANNEL_TRAINING, and container paths. This selection concerns sagemaker script-mode data channel mismatch rather than the required diagnostic-and-remedy pair for rmse versus mae outlier sensitivity. Use evidence: Inspect squared versus absolute residual contributions, target distribution, and large-error business costs. and the matching correction: Compare RMSE with MAE or robust error summaries and choose the metric matching the application's penalty for large misses. Signal: A house-price model's RMSE increases sharply after a few exceptionally expensive properties enter the validation set, although typical absolute errors remain stable. Operational setting: During production evaluation of a SageMaker AI model candidate.
Question 1104
At a production AWS ML data lake supporting customer-facing predictions, the following condition is observed: A demand model treats hour 23 and hour 0 as far apart when integer hour is used as a linear continuous feature. Engineers must avoid changing unrelated source partitions. Which measured outcome would confirm that the correction worked as intended?
- Every held-out forecast uses only data that existed at prediction time and its backtest follows real ordering.
- Replaying the same notification leaves exactly one corresponding logical training record and records deduplication.
- Training converges more consistently, and the same raw request gives aligned standardized values online.
- Neighboring times across midnight map close in cyclic feature space and validation improves without leakage.
Explanation
Correct Answer
D:Neighboring times across midnight map close in cyclic feature space and validation improves without leakage. This is the acceptance evidence for the described time-series cyclical calendar feature scenario. Time-of-day has periodic topology; raw integer distance does not represent wraparound. The corrective implementation it validates is: Encode hour with sine and cosine cyclic features and validate with a time-aware split. Reported signal: A demand model treats hour 23 and hour 0 as far apart when integer hour is used as a linear continuous feature. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
Incorrect Answers
A:Every held-out forecast uses only data that existed at prediction time and its backtest follows real ordering. This could validate chronological split for forecast evaluation, but it does not establish that time-series cyclical calendar feature is resolved for this workload. Confirm instead: Neighboring times across midnight map close in cyclic feature space and validation improves without leakage. Observed problem: A demand model treats hour 23 and hour 0 as far apart when integer hour is used as a linear continuous feature. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
B:Replaying the same notification leaves exactly one corresponding logical training record and records deduplication. This could validate s3 object event at-least-once processing, but it does not establish that time-series cyclical calendar feature is resolved for this workload. Confirm instead: Neighboring times across midnight map close in cyclic feature space and validation improves without leakage. Observed problem: A demand model treats hour 23 and hour 0 as far apart when integer hour is used as a linear continuous feature. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
C:Training converges more consistently, and the same raw request gives aligned standardized values online. This could validate standardization of numerical features, but it does not establish that time-series cyclical calendar feature is resolved for this workload. Confirm instead: Neighboring times across midnight map close in cyclic feature space and validation improves without leakage. Observed problem: A demand model treats hour 23 and hour 0 as far apart when integer hour is used as a linear continuous feature. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
Question 1105
While testing a second region for recoverable deployment of the same ML service, the following condition is observed: A small CPU inference model receives only occasional requests, but a permanently provisioned endpoint stays idle for most hours. The deployment must be reproducible without manual console-only fixes. What targeted change should be implemented after verifying the stated cause?
- Keep the blue environment available until green health gates pass and automate traffic rollback on failed alarms.
- Set adequate minimum capacity, scaling limits, alarms, and request backpressure for the expected traffic ramp.
- Compile the supported model for the specific edge target using SageMaker Neo or choose a compatible conversion route.
- Use SageMaker Serverless Inference where the model and latency envelope fit, or justify provisioned capacity if it does not.
Explanation
Correct Answer
D:Use SageMaker Serverless Inference where the model and latency envelope fit, or justify provisioned capacity if it does not. This is the appropriate controlled remedy for serverless inference for intermittent requests. Serverless capacity trades some cold-start behavior and resource limits for reduced idle endpoint commitment. Verify the cause with: Profile request arrival intervals, cold-start tolerance, memory needs, and compatible serverless inference limits. Then confirm: Low-frequency requests succeed within agreed latency while unused provisioned compute spend is reduced. Reported signal: A small CPU inference model receives only occasional requests, but a permanently provisioned endpoint stays idle for most hours. Operational setting: While testing a second region for recoverable deployment of the same ML service.
Incorrect Answers
A:Keep the blue environment available until green health gates pass and automate traffic rollback on failed alarms. This addresses a different issue involving blue-green endpoint rollback after health failure. It does not correct the stated serverless inference for intermittent requests mechanism. The targeted response is: Use SageMaker Serverless Inference where the model and latency envelope fit, or justify provisioned capacity if it does not. Case evidence: A small CPU inference model receives only occasional requests, but a permanently provisioned endpoint stays idle for most hours. Operational setting: While testing a second region for recoverable deployment of the same ML service.
B:Set adequate minimum capacity, scaling limits, alarms, and request backpressure for the expected traffic ramp. This addresses a different issue involving endpoint scale-to-zero cold capacity planning. It does not correct the stated serverless inference for intermittent requests mechanism. The targeted response is: Use SageMaker Serverless Inference where the model and latency envelope fit, or justify provisioned capacity if it does not. Case evidence: A small CPU inference model receives only occasional requests, but a permanently provisioned endpoint stays idle for most hours. Operational setting: While testing a second region for recoverable deployment of the same ML service.
C:Compile the supported model for the specific edge target using SageMaker Neo or choose a compatible conversion route. This addresses a different issue involving sagemaker neo edge compiler compatibility. It does not correct the stated serverless inference for intermittent requests mechanism. The targeted response is: Use SageMaker Serverless Inference where the model and latency envelope fit, or justify provisioned capacity if it does not. Case evidence: A small CPU inference model receives only occasional requests, but a permanently provisioned endpoint stays idle for most hours. Operational setting: While testing a second region for recoverable deployment of the same ML service.
Question 1106
At the model-governance gate before promotion of a new ML artifact, the following condition is observed: An expensive long-running training job on managed Spot Instances restarts from epoch zero after an interruption because it writes checkpoints only to ephemeral local files. The team needs a defensible technical decision rather than an untracked workaround. What postchange observation provides the strongest acceptance evidence?
- The chosen version meets task metrics and license conditions and its inference contract is reproducible.
- A forced interruption resumes near the last saved epoch and restores optimizer and model state.
- Trials record comparable objective values and a selected candidate meets an independent validation target within budget.
- The tested alert rate and precision match the production workload rather than an optimistic single ROC figure.
Explanation
Correct Answer
B:A forced interruption resumes near the last saved epoch and restores optimizer and model state. This is the acceptance evidence for the described managed spot training checkpoint recovery scenario. Spot capacity can be interrupted; durable checkpoints prevent full retraining when restoration is correctly implemented. The corrective implementation it validates is: Configure a supported SageMaker checkpoint location and resume training state on managed Spot interruption recovery. Reported signal: An expensive long-running training job on managed Spot Instances restarts from epoch zero after an interruption because it writes checkpoints only to ephemeral local files. Operational setting: At the model-governance gate before promotion of a new ML artifact.
Incorrect Answers
A:The chosen version meets task metrics and license conditions and its inference contract is reproducible. This could validate sagemaker jumpstart template provenance, but it does not establish that managed spot training checkpoint recovery is resolved for this workload. Confirm instead: A forced interruption resumes near the last saved epoch and restores optimizer and model state. Observed problem: An expensive long-running training job on managed Spot Instances restarts from epoch zero after an interruption because it writes checkpoints only to ephemeral local files. Operational setting: At the model-governance gate before promotion of a new ML artifact.
C:Trials record comparable objective values and a selected candidate meets an independent validation target within budget. This could validate hyperparameter bayesian tuning strategy, but it does not establish that managed spot training checkpoint recovery is resolved for this workload. Confirm instead: A forced interruption resumes near the last saved epoch and restores optimizer and model state. Observed problem: An expensive long-running training job on managed Spot Instances restarts from epoch zero after an interruption because it writes checkpoints only to ephemeral local files. Operational setting: At the model-governance gate before promotion of a new ML artifact.
D:The tested alert rate and precision match the production workload rather than an optimistic single ROC figure. This could validate roc-auc versus pr-auc for rare events, but it does not establish that managed spot training checkpoint recovery is resolved for this workload. Confirm instead: A forced interruption resumes near the last saved epoch and restores optimizer and model state. Observed problem: An expensive long-running training job on managed Spot Instances restarts from epoch zero after an interruption because it writes checkpoints only to ephemeral local files. Operational setting: At the model-governance gate before promotion of a new ML artifact.
Question 1107
During a controlled replay of the data-preparation workflow in a separate ML staging account, the following condition is observed: An offline churn dataset uses the latest cancellation-status feature for all historical rows, exposing outcomes not known when earlier predictions would run. The dataset contract must remain reproducible for a backfill. Which verification result BEST demonstrates that the issue is resolved?
- Two ordinary runs ingest only newly eligible objects while a separately invoked backfill handles prior intervals.
- For every training row the joined feature timestamp is at or before its prediction timestamp and no label-derived value leaks.
- Neighboring times across midnight map close in cyclic feature space and validation improves without leakage.
- Annotation workers can label required content without viewing direct identifiers and authorized audits can trace approved mappings.
Explanation
Correct Answer
B:For every training row the joined feature timestamp is at or before its prediction timestamp and no label-derived value leaks. This is the acceptance evidence for the described point-in-time feature data leakage scenario. A feature created after the prediction event cannot legitimately inform that earlier prediction. The corrective implementation it validates is: Perform point-in-time correct joins using the most recent eligible pre-prediction feature record, not the globally latest value. Reported signal: An offline churn dataset uses the latest cancellation-status feature for all historical rows, exposing outcomes not known when earlier predictions would run. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
Incorrect Answers
A:Two ordinary runs ingest only newly eligible objects while a separately invoked backfill handles prior intervals. This could validate glue job bookmark state error, but it does not establish that point-in-time feature data leakage is resolved for this workload. Confirm instead: For every training row the joined feature timestamp is at or before its prediction timestamp and no label-derived value leaks. Observed problem: An offline churn dataset uses the latest cancellation-status feature for all historical rows, exposing outcomes not known when earlier predictions would run. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
C:Neighboring times across midnight map close in cyclic feature space and validation improves without leakage. This could validate time-series cyclical calendar feature, but it does not establish that point-in-time feature data leakage is resolved for this workload. Confirm instead: For every training row the joined feature timestamp is at or before its prediction timestamp and no label-derived value leaks. Observed problem: An offline churn dataset uses the latest cancellation-status feature for all historical rows, exposing outcomes not known when earlier predictions would run. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
D:Annotation workers can label required content without viewing direct identifiers and authorized audits can trace approved mappings. This could validate pii masking before model labeling, but it does not establish that point-in-time feature data leakage is resolved for this workload. Confirm instead: For every training row the joined feature timestamp is at or before its prediction timestamp and no label-derived value leaks. Observed problem: An offline churn dataset uses the latest cancellation-status feature for all historical rows, exposing outcomes not known when earlier predictions would run. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
Question 1108
While reproducing the alert in an isolated AWS ML integration account, the following condition is observed: A long training workflow uses only On-Demand capacity even though checkpoints and flexible completion time permit interruptions, and its compute cost breaches target. The proposed change must preserve logging and least privilege. Which verification result BEST demonstrates that the issue is resolved?
- Peak-window 5xx rate and p95 latency fall below service objectives with verified alarms.
- Measured training cost drops while interrupted test jobs resume from checkpoints within the delivery window.
- The new container operates with rotated credentials and image scans find no active embedded password.
- A test write to an approved prefix succeeds, while writes to unapproved external destinations are denied.
Explanation
Correct Answer
B:Measured training cost drops while interrupted test jobs resume from checkpoints within the delivery window. This is the acceptance evidence for the described spot training cost versus interruption recovery scenario. Spot savings depend on interruption handling; a lower hourly rate is not a guarantee of lower total job cost. The corrective implementation it validates is: Use supported managed Spot Training with durable checkpoints for interruptible phases and retain On-Demand capacity when deadlines demand it. Reported signal: A long training workflow uses only On-Demand capacity even though checkpoints and flexible completion time permit interruptions, and its compute cost breaches target. Operational setting: While reproducing the alert in an isolated AWS ML integration account.
Incorrect Answers
A:Peak-window 5xx rate and p95 latency fall below service objectives with verified alarms. This could validate cloudwatch 5xxerrors endpoint saturation, but it does not establish that spot training cost versus interruption recovery is resolved for this workload. Confirm instead: Measured training cost drops while interrupted test jobs resume from checkpoints within the delivery window. Observed problem: A long training workflow uses only On-Demand capacity even though checkpoints and flexible completion time permit interruptions, and its compute cost breaches target. Operational setting: While reproducing the alert in an isolated AWS ML integration account.
C:The new container operates with rotated credentials and image scans find no active embedded password. This could validate secret rotation for inference container credentials, but it does not establish that spot training cost versus interruption recovery is resolved for this workload. Confirm instead: Measured training cost drops while interrupted test jobs resume from checkpoints within the delivery window. Observed problem: A long training workflow uses only On-Demand capacity even though checkpoints and flexible completion time permit interruptions, and its compute cost breaches target. Operational setting: While reproducing the alert in an isolated AWS ML integration account.
D:A test write to an approved prefix succeeds, while writes to unapproved external destinations are denied. This could validate cross-account s3 bucket exfiltration protection, but it does not establish that spot training cost versus interruption recovery is resolved for this workload. Confirm instead: Measured training cost drops while interrupted test jobs resume from checkpoints within the delivery window. Observed problem: A long training workflow uses only On-Demand capacity even though checkpoints and flexible completion time permit interruptions, and its compute cost breaches target. Operational setting: While reproducing the alert in an isolated AWS ML integration account.
Question 1109
At a production AWS ML data lake supporting customer-facing predictions, the following condition is observed: A hiring-model dataset has very different positive-label rates across sensitive groups, but the team compares only global accuracy. Engineers must avoid changing unrelated source partitions. Which investigation provides the most decisive FIRST evidence for the underlying problem?
- Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report.
- Measure file-size distribution, object request count, Spark task scheduling overhead, and read throughput.
- Inspect shared entity identifiers across partitions and compare patient-level versus image-level evaluation results.
- Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers.
Explanation
Correct Answer
A:Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report. This directly tests clarify label imbalance measurement in the reported AWS ML workflow. Aggregate accuracy can conceal unequal source-label distributions, which pretraining metrics assess before model fitting. It distinguishes the problem before choosing a change. The related targeted response is: Measure label distribution differences with an appropriate pretraining bias metric and improve data representation before training. Reported signal: A hiring-model dataset has very different positive-label rates across sensitive groups, but the team compares only global accuracy. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
Incorrect Answers
B:Measure file-size distribution, object request count, Spark task scheduling overhead, and read throughput. This is a possible diagnostic for columnar file compaction for small objects, but it does not directly test the clarify label imbalance measurement failure reported here. Prioritize: Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report. The relevant case is: A hiring-model dataset has very different positive-label rates across sensitive groups, but the team compares only global accuracy. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
C:Inspect shared entity identifiers across partitions and compare patient-level versus image-level evaluation results. This is a possible diagnostic for group-leakage train and validation split, but it does not directly test the clarify label imbalance measurement failure reported here. Prioritize: Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report. The relevant case is: A hiring-model dataset has very different positive-label rates across sensitive groups, but the team compares only global accuracy. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
D:Inspect categorical cardinality, sparse matrix dimensions, feature contribution, and generalization to unseen identifiers. This is a possible diagnostic for high-cardinality categorical encoding, but it does not directly test the clarify label imbalance measurement failure reported here. Prioritize: Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report. The relevant case is: A hiring-model dataset has very different positive-label rates across sensitive groups, but the team compares only global accuracy. Operational setting: At a production AWS ML data lake supporting customer-facing predictions.
Question 1110
During a controlled replay of the data-preparation workflow in a separate ML staging account, the following condition is observed: New event producers change a numerical source field to strings while legacy objects remain numeric, causing the shared Glue transformation to fail. The dataset contract must remain reproducible for a backfill. What should the ML engineer inspect FIRST to distinguish this failure from unrelated issues?
- Compare online Feature Store record values and offline event-time keyed feature history for the same entity and model snapshot.
- Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report.
- Compare source schemas by partition, Glue Data Catalog types, and failed cast rows for the evolving field.
- Check job names, bookmark settings, run history, source object timestamps, and the previously committed checkpoint.
Explanation
Correct Answer
C:Compare source schemas by partition, Glue Data Catalog types, and failed cast rows for the evolving field. This directly tests glue schema evolution with mixed input types in the reported AWS ML workflow. A catalog entry does not automatically reconcile incompatible field types across all source partitions. It distinguishes the problem before choosing a change. The related targeted response is: Define explicit compatible conversions and schema contracts, isolate incompatible versions, and validate before publishing features. Reported signal: New event producers change a numerical source field to strings while legacy objects remain numeric, causing the shared Glue transformation to fail. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
Incorrect Answers
A:Compare online Feature Store record values and offline event-time keyed feature history for the same entity and model snapshot. This is a possible diagnostic for sagemaker feature store online offline consistency, but it does not directly test the glue schema evolution with mixed input types failure reported here. Prioritize: Compare source schemas by partition, Glue Data Catalog types, and failed cast rows for the evolving field. The relevant case is: New event producers change a numerical source field to strings while legacy objects remain numeric, causing the shared Glue transformation to fail. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
B:Inspect group sample sizes, label-positive fractions, and the appropriate SageMaker Clarify pretraining bias report. This is a possible diagnostic for clarify label imbalance measurement, but it does not directly test the glue schema evolution with mixed input types failure reported here. Prioritize: Compare source schemas by partition, Glue Data Catalog types, and failed cast rows for the evolving field. The relevant case is: New event producers change a numerical source field to strings while legacy objects remain numeric, causing the shared Glue transformation to fail. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
D:Check job names, bookmark settings, run history, source object timestamps, and the previously committed checkpoint. This is a possible diagnostic for glue job bookmark state error, but it does not directly test the glue schema evolution with mixed input types failure reported here. Prioritize: Compare source schemas by partition, Glue Data Catalog types, and failed cast rows for the evolving field. The relevant case is: New event producers change a numerical source field to strings while legacy objects remain numeric, causing the shared Glue transformation to fail. Operational setting: During a controlled replay of the data-preparation workflow in a separate ML staging account.
