Exam Bundle

A multilingual test phrase yields the same token IDs in the training pipeline and deployed inference endpoint. This would verify a different operation: Version a single canonical text normalization and tokenizer artifact and use it in both processing and serving containers. That separate action addresses: A text classifier tokenizes raw Unicode text differently in batch training and the real-time inference application. It cannot establish the required Timberline Foods correction for: A time-series demand model shuffles future demand records into an earlier training window, invalidating its forward-looking evaluation. The decisive acceptance check is instead: The training cutoff precedes the first validation observation and a forward rolling evaluation reproduces the expected boundaries.

Model input checks confirm no implied numerical ordering and all expected regions map to the approved encoded features. This would verify a different operation: Apply an appropriate categorical encoding while preserving the region identifier's nominal semantics and missing-value policy. That separate action addresses: A numeric categorical code such as region ID is mistakenly treated as an ordered magnitude when supplied to a linear model. It cannot establish the required Timberline Foods correction for: A time-series demand model shuffles future demand records into an earlier training window, invalidating its forward-looking evaluation. The decisive acceptance check is instead: The training cutoff precedes the first validation observation and a forward rolling evaluation reproduces the expected boundaries.

A sampled reconciliation shows expected one-to-many relationships and the final training row count meets the documented bounds. This would verify a different operation: Define an authoritative join key and a validated join cardinality before merging the two approved data sources. That separate action addresses: An ML team combines order data from RDS and session data from S3 without a stable customer join key and accidentally multiplies training rows. It cannot establish the required Timberline Foods correction for: A time-series demand model shuffles future demand records into an earlier training window, invalidating its forward-looking evaluation. The decisive acceptance check is instead: The training cutoff precedes the first validation observation and a forward rolling evaluation reproduces the expected boundaries.

Materialize governed versioned features with an appropriate offline store and documented recomputation schedule. It does not directly address the Westhaven Retail case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Implement the same window boundaries, event-time cutoff and aggregation logic for offline and online feature creation. It does not directly address the Westhaven Retail case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Adopt a tested document chunking and aggregation strategy aligned to the model's maximum input length. It does not directly address the Westhaven Retail case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Subsequent readers select the newly committed version and no partial object contents are exposed as an accepted input. This would verify a different operation: Change the workflow to write complete immutable S3 objects and publish a new object or manifest version instead of POSIX-style append. That separate action addresses: A data ingestion script assumes S3 is a POSIX filesystem and tries in-place append operations on one object during batch preparation. It cannot establish the required Westhaven Retail correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

A reproducible export contains the required records while production capacity and latency remain inside agreed limits. This would verify a different operation: Use a designed DynamoDB export to S3 or appropriate controlled query pattern instead of repeated unrestricted production table scans. That separate action addresses: A training pipeline needs recent transaction attributes from DynamoDB, but its ad hoc full table scan increases read capacity consumption during business hours. It cannot establish the required Westhaven Retail correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Distance-sensitive evaluation improves and the identical stored scale parameters are used during training and online inference. This would verify a different operation: Fit a standardization or scaling transform on training data only and apply the saved transform to validation and inference inputs. That separate action addresses: A churn model trains on customer age and transaction amount with unrelated scales, causing its distance-based feature representation to be dominated by transaction magnitude. It cannot establish the required Westhaven Retail correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Deduplicate on stable business identity and event-version rules, retaining the latest accepted entity state as required. It does not directly address the Granite SaaS case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Perform approved irreversible masking or tokenization before feature extraction and restrict the original identifier fields. It does not directly address the Granite SaaS case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Measure partition-key distribution and provision or scale stream capacity using suitable shard or on-demand settings. It does not directly address the Granite SaaS case: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The relevant acceptance criterion for this case is: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Two deliveries of one test event result in one accepted logical feature update and a failed payload remains retriable. This would verify a different operation: Make the feature write idempotent using event identifiers or version checks and send repeated failures to a controlled recovery path. That separate action addresses: A streaming Lambda transformer retries failed events but appends duplicate feature writes each time its invocation is replayed. It cannot establish the required Granite SaaS correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

The feature vector dimensions stay bounded and an evaluation report measures collisions or embedding quality on held-out accounts. This would verify a different operation: Use a suitable embedding or hashing approach with collision and memory impact assessed for the model objective. That separate action addresses: A recommendation model's high-cardinality account IDs explode into impractically wide one-hot sparse features. It cannot establish the required Granite SaaS correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

A second training run reads the expected materialized feature version and shows lower repeated transformation time. This would verify a different operation: Materialize governed versioned features with an appropriate offline store and documented recomputation schedule. That separate action addresses: A tabular feature set repeatedly recomputes the same expensive historical aggregations for every training job without a reusable feature definition. It cannot establish the required Granite SaaS correction for: A classification training set contains duplicated near-identical product images assigned conflicting labels, creating avoidable noise. The decisive acceptance check is instead: The duplicate audit shows coherent labels within matching-image clusters and a reviewed exception log tracks ambiguous cases.

Fit and persist a categorical encoder on the training set with an explicit unknown-category handling policy for all later inputs. It does not directly address the Bluehaven Transit case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Grant narrowly scoped source-object and KMS decrypt permissions to the authorized consuming role and confirm the key policy permits the operation. It does not directly address the Bluehaven Transit case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Align the SageMaker processing input URI with the actual produced S3 dataset location and validate the execution role's read permission. It does not directly address the Bluehaven Transit case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Zero and positive test values transform into finite documented values and the resulting feature remains suitable for the model. This would verify a different operation: Choose a guarded nonnegative log transformation and test boundary handling before saving the feature pipeline. That separate action addresses: An ML regression dataset has a heavily skewed positive feature, but an unguarded logarithmic transform generates invalid values for zero entries. It cannot establish the required Bluehaven Transit correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

The controlled high-rate producer stays within accepted write limits and downstream lag returns to the target threshold. This would verify a different operation: Measure partition-key distribution and provision or scale stream capacity using suitable shard or on-demand settings. That separate action addresses: A high-volume inference feature ingest uses one Kinesis shard even after throughput grows, leading to sustained throttling of producer records. It cannot establish the required Bluehaven Transit correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

The training schema references the intended version and a sampled stored record matches its declared transformation lineage. This would verify a different operation: Use explicit versioned feature names and document producer-consumer contracts for the stored feature definitions. That separate action addresses: A SageMaker Feature Store group publishes a transformed feature but training continues to read an older raw column with the same business label. It cannot establish the required Bluehaven Transit correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Apply an appropriate categorical encoding while preserving the region identifier's nominal semantics and missing-value policy. It does not directly address the Brookside Manufacturing case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Implement the same window boundaries, event-time cutoff and aggregation logic for offline and online feature creation. It does not directly address the Brookside Manufacturing case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Register and validate an explicit schema or compatible evolution policy before promoting events into the curated training zone. It does not directly address the Brookside Manufacturing case: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The relevant acceptance criterion for this case is: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

A deliberately delayed record lands in its documented event-time window rather than the time when the collector received it. This would verify a different operation: Retain event-time timestamps and apply the streaming processor's supported window and watermark policy for late arrivals. That separate action addresses: A streaming feature pipeline stores event timestamps as ingestion time only, causing late-arriving mobile events to be counted in the wrong training interval. It cannot establish the required Brookside Manufacturing correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Model input checks confirm no implied numerical ordering and all expected regions map to the approved encoded features. This would verify a different operation: Apply an appropriate categorical encoding while preserving the region identifier's nominal semantics and missing-value policy. That separate action addresses: A numeric categorical code such as region ID is mistakenly treated as an ordered magnitude when supplied to a linear model. It cannot establish the required Brookside Manufacturing correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

The retained training sample preserves known rare positives and a held-out fraud recall check remains within the target band. This would verify a different operation: Profile the feature distribution by label and apply documented context-aware outlier handling rather than global blind filtering. That separate action addresses: A data-wrangling job treats all high-value transactions as outliers and deletes legitimate rare fraud cases critical to model recall. It cannot establish the required Brookside Manufacturing correction for: SageMaker Clarify pre-training bias metrics flag a demographic proportion imbalance but the team reports only overall training loss. The decisive acceptance check is instead: The baseline report contains relevant protected-group comparisons and an accountable review documents any mitigation and residual risk.

Define a validated label encoding dictionary and reject unmapped target values before model training starts. It does not directly address the Harborline Logistics case: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The relevant acceptance criterion for this case is: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.

Use a suitable embedding or hashing approach with collision and memory impact assessed for the model objective. It does not directly address the Harborline Logistics case: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The relevant acceptance criterion for this case is: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.

Select scalable shared or object storage such as S3 or compatible FSx for Lustre access according to the actual training access pattern. It does not directly address the Harborline Logistics case: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The relevant acceptance criterion for this case is: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.

Two deliveries of one test event result in one accepted logical feature update and a failed payload remains retriable. This would verify a different operation: Make the feature write idempotent using event identifiers or version checks and send repeated failures to a controlled recovery path. That separate action addresses: A streaming Lambda transformer retries failed events but appends duplicate feature writes each time its invocation is replayed. It cannot establish the required Harborline Logistics correction for: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The decisive acceptance check is instead: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.

A second training run reads the expected materialized feature version and shows lower repeated transformation time. This would verify a different operation: Materialize governed versioned features with an appropriate offline store and documented recomputation schedule. That separate action addresses: A tabular feature set repeatedly recomputes the same expensive historical aggregations for every training job without a reusable feature definition. It cannot establish the required Harborline Logistics correction for: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The decisive acceptance check is instead: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.

The large synthetic partition completes without driver out-of-memory errors and its aggregate counts match the reference run. This would verify a different operation: Use distributed Spark transforms or scalable partitioned outputs instead of collecting the full dataset to one process. That separate action addresses: A Spark feature preparation stage converts all encoded records to driver memory and fails for large training partitions. It cannot establish the required Harborline Logistics correction for: A training manifest refers to S3 objects that were removed by an aggressive lifecycle rule, so reproducible model builds fail months later. The decisive acceptance check is instead: A scheduled retention test can still retrieve every manifest object needed to reproduce the archived model version.