Google Professional Machine Learning Engineer Practice Test 105
Question 1040
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A Cloud Storage finalize event triggers training for every object, including temporary files, causing duplicate and premature model runs. The team requires an idempotent pipeline behavior and traceable evidence. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger.
- Inspect Cloud Build substitutions, commit SHAs, model registry versions, signed image digests and approval records.
- Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs.
- Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
- Create immutable versioned input snapshots or partition manifests and record them in training metadata.
Explanation
Correct Answers
A:Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger. This is the targeted corrective response for eventbridge equivalent event trigger on storage. Storage event notifications require business-level readiness and deduplication before training begins. Verify the diagnosis using: Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs. Confirm success using: One completed batch starts one intended pipeline run and unrelated object uploads do not trigger training. This matters here because while reproducing a training orchestration incident in isolated staging, the following condition occurs: A Cloud….
C:Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs. This inspects evidence relevant to eventbridge equivalent event trigger on storage. Storage event notifications require business-level readiness and deduplication before training begins. After identifying the cause, the matching correction is: Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger. This matters here because while reproducing a training orchestration incident in isolated staging, the following condition occurs: A Cloud….
Incorrect Answers
B:Inspect Cloud Build substitutions, commit SHAs, model registry versions, signed image digests and approval records. This is an action or diagnostic relevant to cloud build release version promotion, not one of the required two steps. In this scenario verify: Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs. Then correct: Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger.
D:Publish one versioned pipeline package and promote it with reviewed environment-specific parameters. This is an action or diagnostic relevant to pipeline environment parameter promotion, not one of the required two steps. In this scenario verify: Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs. Then correct: Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger.
E:Create immutable versioned input snapshots or partition manifests and record them in training metadata. This is an action or diagnostic relevant to data version reproducibility in ct jobs, not one of the required two steps. In this scenario verify: Inspect object prefixes, event filters, completed-batch manifests and recent pipeline execution IDs. Then correct: Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger. This matters here because while reproducing a training orchestration incident in isolated staging, the following condition occurs: A Cloud….
Question 1041
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A model registry entry lacks the upstream version of the training feature table, preventing investigators from reconstructing a reported prediction. The team requires an idempotent pipeline behavior and traceable evidence. Which configuration or procedure is the most appropriate correction?
- Fail protected build or release stages when prohibited vulnerabilities exist and require documented exceptions for approved deviations.
- Trigger candidate retraining only when justified evidence and sufficient labeled data are available, with evaluation gates.
- Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact.
- Use supported Ray retry policies and durable checkpoints at suitable stage boundaries instead of relying only on worker memory.
Explanation
Correct Answer
C:Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact. This is the targeted corrective response for model feature lineage for audit. Model weights alone do not capture the data lineage necessary to reproduce and investigate predictions. Verify the diagnosis using: Inspect ML Metadata execution links, table snapshots, transformations and model version lineage. Confirm success using: An audit can resolve model version to exact source data and processing code without guessing by timestamps.
Incorrect Answers
A:Fail protected build or release stages when prohibited vulnerabilities exist and require documented exceptions for approved deviations. This action addresses cloud build container security gate, not the failure in the described model feature lineage for audit case. The targeted correction is: Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact.
B:Trigger candidate retraining only when justified evidence and sufficient labeled data are available, with evaluation gates. This action addresses data drift triggered retraining policy, not the failure in the described model feature lineage for audit case. The targeted correction is: Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact.
D:Use supported Ray retry policies and durable checkpoints at suitable stage boundaries instead of relying only on worker memory. This action addresses ray distributed task failure isolation, not the failure in the described model feature lineage for audit case. The targeted correction is: Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact.
Question 1042
During model-training validation on a Google Cloud experimental project, the following condition occurs: A team wants to tune an image-generation foundation model using scraped brand imagery without verifying usage rights or consent. The change should preserve a meaningful independent model-quality comparison. Which diagnostic check should be performed FIRST to isolate the described cause?
- Compare sampler configuration, example IDs, per-worker output counts and the global metric-reduction step.
- Check dataset licenses, provenance, consent policies, safety requirements and the model's supported tuning route.
- Inspect task input configuration, container logs, expected file patterns and staged object paths.
- Examine model governance constraints, feature sensitivity, explanation fidelity and performance of simpler interpretable baselines.
Explanation
Correct Answer
B:Check dataset licenses, provenance, consent policies, safety requirements and the model's supported tuning route. This inspects evidence relevant to fine-tuning image generator data rights. Fine-tuning technical feasibility does not confer rights to reuse restricted training images or ignore model safety. After identifying the cause, the matching correction is: Use licensed and permitted training images with appropriate governance reviews before any supported tuning job. This matters here because during model-training validation on a Google Cloud experimental project, the following condition occurs: A team….
Incorrect Answers
A:Compare sampler configuration, example IDs, per-worker output counts and the global metric-reduction step. This examines data-parallel evaluation duplicate samples rather than the specific mechanism reported here. It is not the decisive first check for fine-tuning image generator data rights. The relevant evidence is: Check dataset licenses, provenance, consent policies, safety requirements and the model's supported tuning route.
C:Inspect task input configuration, container logs, expected file patterns and staged object paths. This examines custom training input schema mismatch rather than the specific mechanism reported here. It is not the decisive first check for fine-tuning image generator data rights. The relevant evidence is: Check dataset licenses, provenance, consent policies, safety requirements and the model's supported tuning route.
D:Examine model governance constraints, feature sensitivity, explanation fidelity and performance of simpler interpretable baselines. This examines interpretability-constrained model baseline rather than the specific mechanism reported here. It is not the decisive first check for fine-tuning image generator data rights. The relevant evidence is: Check dataset licenses, provenance, consent policies, safety requirements and the model's supported tuning route.
Question 1043
During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: Two ML engineers rerun the same notebook and obtain incompatible model artifacts because one environment silently installs a newer Python library release. The handoff must preserve privacy, lineage and testable behavior. What implementation change resolves the specific issue without substituting an unrelated fix?
- Quarantine invalid spikes and fit a robust transformation using training data only, then apply it consistently at inference.
- Choose a compatible Model Garden model/runtime or perform a validated checkpoint conversion before training.
- Refactor hidden state into explicit inputs and functions and add a clean-kernel end-to-end execution gate.
- Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes.
Explanation
Correct Answer
D:Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes. This is the targeted corrective response for notebook dependency lock reproducibility. Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment. Verify the diagnosis using: Inspect Python runtime versions, pip dependency trees, package hashes and notebook execution metadata for both runs. Confirm success using: Independent notebook runs resolve the same approved libraries and produce comparable metric and artifact outputs.
Incorrect Answers
A:Quarantine invalid spikes and fit a robust transformation using training data only, then apply it consistently at inference. This action addresses outlier handling before feature standardization, not the failure in the described notebook dependency lock reproducibility case. The targeted correction is: Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes.
B:Choose a compatible Model Garden model/runtime or perform a validated checkpoint conversion before training. This action addresses model garden jax accelerator compatibility, not the failure in the described notebook dependency lock reproducibility case. The targeted correction is: Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes.
C:Refactor hidden state into explicit inputs and functions and add a clean-kernel end-to-end execution gate. This action addresses notebook transient state and execution order, not the failure in the described notebook dependency lock reproducibility case. The targeted correction is: Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes.
Question 1044
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A scheduled training pipeline fails during feature validation but the team discovers the failure only when customers complain about stale predictions. The team requires an idempotent pipeline behavior and traceable evidence. Select TWO: the decisive diagnostic step and the targeted repair for this specific issue.
- Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest.
- Make the load idempotent with keyed merge or overwrite-per-version semantics and safe retry handling.
- Promote explicit immutable model and image versions with versioned release manifests and review gates.
- Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings.
- Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives.
Explanation
Correct Answers
D:Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. This inspects evidence relevant to pipeline notification alert on silent failure. Orchestration without failure notification can leave models stale even when the last successful deployment was correct. After identifying the cause, the matching correction is: Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives. This matters here because while reproducing a training orchestration incident in isolated staging, the following condition occurs: A scheduled….
E:Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives. This is the targeted corrective response for pipeline notification alert on silent failure. Orchestration without failure notification can leave models stale even when the last successful deployment was correct. Verify the diagnosis using: Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. Confirm success using: An injected validation failure alerts the responsible owner and the last healthy model continues serving. This matters here because while reproducing a training orchestration incident in isolated staging, the following condition occurs: A scheduled….
Incorrect Answers
A:Inspect snapshot identifiers, SQL filters, table partitions and the model version's recorded input manifest. This is an action or diagnostic relevant to data version reproducibility in ct jobs, not one of the required two steps. In this scenario verify: Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. Then correct: Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives.
B:Make the load idempotent with keyed merge or overwrite-per-version semantics and safe retry handling. This is an action or diagnostic relevant to airflow task idempotency on retry, not one of the required two steps. In this scenario verify: Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. Then correct: Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives.
C:Promote explicit immutable model and image versions with versioned release manifests and review gates. This is an action or diagnostic relevant to cloud build release version promotion, not one of the required two steps. In this scenario verify: Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. Then correct: Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives.
Question 1045
During a regulated MLOps pipeline deployment, the following condition occurs: A production workflow starts model registration before validation and evaluation steps have finished, occasionally promoting untested weights. The workflow must be repeatable and protect the last approved production model. Which targeted engineering change most directly addresses the failure?
- Add explicit upstream validation and evaluation dependencies before registration and deployment components.
- Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
- Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger.
- Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
Explanation
Correct Answer
A:Add explicit upstream validation and evaluation dependencies before registration and deployment components. This is the targeted corrective response for pipeline component dependency order. Orchestration needs explicit data and control dependencies; notebook execution order is not a production DAG. Verify the diagnosis using: Inspect the pipeline DAG, dependency declarations, artifact references and execution ordering from recent runs. Confirm success using: Every production model version traces to successfully completed validation and evaluation steps in that run.
Incorrect Answers
B:Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching. This action addresses kubeflow component cache invalidation, not the failure in the described pipeline component dependency order case. The targeted correction is: Add explicit upstream validation and evaluation dependencies before registration and deployment components.
C:Filter validated completed-dataset events and enforce an idempotent manifest-based training trigger. This action addresses eventbridge equivalent event trigger on storage, not the failure in the described pipeline component dependency order case. The targeted correction is: Add explicit upstream validation and evaluation dependencies before registration and deployment components.
D:Enforce a shared typed schema contract and reject incompatible feature versions before training or inference. This action addresses serving schema validation component, not the failure in the described pipeline component dependency order case. The targeted correction is: Add explicit upstream validation and evaluation dependencies before registration and deployment components.
Question 1046
During a staged restoration test at a secondary serving deployment, the following condition occurs: A multi-class model returns numeric class indices, but an API gateway uses a stale alphabetical mapping and mislabels predictions. The team must prove an actual user-visible improvement without weakening access controls. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Benchmark a Cloud Run custom container against a managed Agent Platform endpoint and choose for workload fit.
- Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency.
- Ship the class vocabulary with the model artifact and version or validate the output adapter against it.
- Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version.
- Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.
Explanation
Correct Answers
C:Ship the class vocabulary with the model artifact and version or validate the output adapter against it. This is the targeted corrective response for custom postprocessing label mapping drift. Index ordering is part of the model contract; separate stale label dictionaries can silently corrupt inference. Verify the diagnosis using: Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version. Confirm success using: Every controlled sample's returned label matches the model's trained class index with no misordered classes. This matters here because during a staged restoration test at a secondary serving deployment, the following condition occurs: A multi-class….
D:Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version. This inspects evidence relevant to custom postprocessing label mapping drift. Index ordering is part of the model contract; separate stale label dictionaries can silently corrupt inference. After identifying the cause, the matching correction is: Ship the class vocabulary with the model artifact and version or validate the output adapter against it. This matters here because during a staged restoration test at a secondary serving deployment, the following condition occurs: A multi-class….
Incorrect Answers
A:Benchmark a Cloud Run custom container against a managed Agent Platform endpoint and choose for workload fit. This is an action or diagnostic relevant to cloud run versus managed inference endpoint, not one of the required two steps. In this scenario verify: Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version. Then correct: Ship the class vocabulary with the model artifact and version or validate the output adapter against it.
B:Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency. This is an action or diagnostic relevant to dynamic batch serving for gpu saturation, not one of the required two steps. In this scenario verify: Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version. Then correct: Ship the class vocabulary with the model artifact and version or validate the output adapter against it.
E:Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements. This is an action or diagnostic relevant to private endpoint for regulated inference traffic, not one of the required two steps. In this scenario verify: Inspect trained class-to-index metadata, endpoint output tensors and gateway mapping version. Then correct: Ship the class vocabulary with the model artifact and version or validate the output adapter against it.
Question 1047
During model-training validation on a Google Cloud experimental project, the following condition occurs: A custom PyTorch training job runs out of accelerator memory immediately after per-device batch size doubles while model architecture stays fixed. The change should preserve a meaningful independent model-quality comparison. Which configuration or procedure is the most appropriate correction?
- Reduce per-device batch size or use validated gradient accumulation or mixed precision where compatible.
- Recompute lagged windows using only prior observations and validate with rolling-origin splits.
- Split raw records first and resample only inside training folds, leaving validation cohorts untouched.
- Use supported structured output and schema validation with graceful retry or rejection for nonconforming responses.
Explanation
Correct Answer
A:Reduce per-device batch size or use validated gradient accumulation or mixed precision where compatible. This is the targeted corrective response for gpu out-of-memory batch control. Training activation memory commonly increases with batch size, independent of available CPU cores. Verify the diagnosis using: Inspect GPU memory traces, activation tensor sizes, batch size, optimizer state and training failure logs. Confirm success using: The job completes representative epochs without OOM and convergence remains within the approved tolerance. This matters here because during model-training validation on a Google Cloud experimental project, the following condition occurs: A custom….
Incorrect Answers
B:Recompute lagged windows using only prior observations and validate with rolling-origin splits. This action addresses temporal training label leakage in rolling windows, not the failure in the described gpu out-of-memory batch control case. The targeted correction is: Reduce per-device batch size or use validated gradient accumulation or mixed precision where compatible.
C:Split raw records first and resample only inside training folds, leaving validation cohorts untouched. This action addresses training class imbalance fold leakage, not the failure in the described gpu out-of-memory batch control case. The targeted correction is: Reduce per-device batch size or use validated gradient accumulation or mixed precision where compatible.
D:Use supported structured output and schema validation with graceful retry or rejection for nonconforming responses. This action addresses constrained generative json outputs, not the failure in the described gpu out-of-memory batch control case. The targeted correction is: Reduce per-device batch size or use validated gradient accumulation or mixed precision where compatible.
Question 1048
Within a jointly maintained Google Cloud data science workspace, the following condition occurs: A distributed preprocessing job joins two huge feature tables, but one NULL-like customer category contains most rows and one reducer becomes the bottleneck. Collaborators must avoid expanding data access or breaking experiment reproducibility. Which targeted engineering change most directly addresses the failure?
- Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts.
- Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.
- Register a common versioned holdout and comparable metrics in Agent Platform Experiments before ranking candidates.
- Rewrite the filter to permit effective partition pruning while preserving its requested date boundaries.
Explanation
Correct Answer
A:Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts. This is the targeted corrective response for managed spark join-key skew repair. Skew concentrates otherwise distributed work on a single hot key, which extra workers alone may not resolve. Verify the diagnosis using: Inspect join-key frequency distributions, per-task runtime, shuffle spill and the output cardinality before optimization. Confirm success using: Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation.
Incorrect Answers
B:Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment. This action addresses point-in-time feature leakage prevention, not the failure in the described managed spark join-key skew repair case. The targeted correction is: Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts.
C:Register a common versioned holdout and comparable metrics in Agent Platform Experiments before ranking candidates. This action addresses experiment metric comparison with mismatched splits, not the failure in the described managed spark join-key skew repair case. The targeted correction is: Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts.
D:Rewrite the filter to permit effective partition pruning while preserving its requested date boundaries. This action addresses bigquery partition filter cost optimization, not the failure in the described managed spark join-key skew repair case. The targeted correction is: Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts.
Question 1049
Within a jointly maintained Google Cloud data science workspace, the following condition occurs: Two ML engineers rerun the same notebook and obtain incompatible model artifacts because one environment silently installs a newer Python library release. Collaborators must avoid expanding data access or breaking experiment reproducibility. Which underlying engineering principle explains why the original approach fails?
- Long-lived secrets copied into notebooks or Git can be reused outside intended workloads and must be rotated.
- Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment.
- MLOps provenance requires a recorded chain from code and data to model and deployment, not merely a friendly name.
- Framework-native checkpoints and runtime APIs are not automatically interchangeable across ML libraries.
Explanation
Correct Answer
B:Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment. This explains the mechanism behind notebook dependency lock reproducibility and why the described failure is possible. A discriminating investigation is: Inspect Python runtime versions, pip dependency trees, package hashes and notebook execution metadata for both runs. The corresponding repair is: Pin compatible dependency versions in a reproducible environment specification and validate fresh builds on clean runtimes. This matters here because within a jointly maintained Google Cloud data science workspace, the following condition occurs: Two ML engineers….
Incorrect Answers
A:Long-lived secrets copied into notebooks or Git can be reused outside intended workloads and must be rotated. This principle describes workbench credentials embedded in repository, which is not the cause of the observed notebook dependency lock reproducibility failure. The controlling explanation is: Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment.
C:MLOps provenance requires a recorded chain from code and data to model and deployment, not merely a friendly name. This principle describes ml metadata lineage for production model, which is not the cause of the observed notebook dependency lock reproducibility failure. The controlling explanation is: Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment.
D:Framework-native checkpoints and runtime APIs are not automatically interchangeable across ML libraries. This principle describes model garden jax accelerator compatibility, which is not the cause of the observed notebook dependency lock reproducibility failure. The controlling explanation is: Notebook cells alone do not capture interpreter and package state; reproducibility depends on the execution environment. This matters here because within a jointly maintained Google Cloud data science workspace, the following condition occurs: Two ML engineers….
Question 1050
During a production ML inference release on Google Cloud, the following condition occurs: A marketing model scores ten million customers every midnight, but engineers keep a low-latency online endpoint running continuously for the once-daily batch. The service must meet its agreed quality and latency requirements. Which targeted engineering change most directly addresses the failure?
- Implement the required serving contract and make the container report readiness only after loading model assets.
- Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost.
- Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint.
- Ship the class vocabulary with the model artifact and version or validate the output adapter against it.
Explanation
Correct Answer
C:Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint. This is the targeted corrective response for batch versus online inference schedule. Batch workloads need throughput and scheduling guarantees, whereas real-time endpoints serve interactive low-latency calls. Verify the diagnosis using: Examine scoring cadence, input size, end-to-end completion window, idle endpoint cost and per-record response requirements. Confirm success using: The nightly job completes within the agreed SLA without idle real-time serving charges. This matters here because a marketing model scores ten million customers every midnight, but engineers keep a low-latency online endpoint….
Incorrect Answers
A:Implement the required serving contract and make the container report readiness only after loading model assets. This action addresses inference container health interface mismatch, not the failure in the described batch versus online inference schedule case. The targeted correction is: Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint.
B:Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost. This action addresses autoscaling minimum capacity versus cold load, not the failure in the described batch versus online inference schedule case. The targeted correction is: Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint.
D:Ship the class vocabulary with the model artifact and version or validate the output adapter against it. This action addresses custom postprocessing label mapping drift, not the failure in the described batch versus online inference schedule case. The targeted correction is: Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint.