Google Professional Machine Learning Engineer Practice Test 14
Question 130
During a staged restoration test at a secondary serving deployment, the following condition occurs: A team deploys a model using a mutable storage path named production-latest and cannot reconstruct which evaluation approved its weights. The team must prove an actual user-visible improvement without weakening access controls. Which evidence most directly establishes what is wrong with this AI or ML workflow?
- Inspect batch size, kernel utilization, per-request overhead, queue wait and latency percentiles.
- Review immutable artifact IDs, Model Registry versions, evaluation reports, approvals and deployed revision metadata.
- Inspect endpoint revision metrics, release alarms, traffic fractions and availability of the prior known-good model.
- Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
Explanation
Correct Answer
B:Review immutable artifact IDs, Model Registry versions, evaluation reports, approvals and deployed revision metadata. This inspects evidence relevant to model registry version provenance. Mutable nicknames do not establish immutable model provenance for controlled production releases. After identifying the cause, the matching correction is: Promote an explicitly approved model version from Agent Platform Model Registry and record deployment lineage.
Incorrect Answers
A:Inspect batch size, kernel utilization, per-request overhead, queue wait and latency percentiles. This examines dynamic batch serving for gpu saturation rather than the specific mechanism reported here. It is not the decisive first check for model registry version provenance. The relevant evidence is: Review immutable artifact IDs, Model Registry versions, evaluation reports, approvals and deployed revision metadata.
C:Inspect endpoint revision metrics, release alarms, traffic fractions and availability of the prior known-good model. This examines model traffic rollback on high error rate rather than the specific mechanism reported here. It is not the decisive first check for model registry version provenance. The relevant evidence is: Review immutable artifact IDs, Model Registry versions, evaluation reports, approvals and deployed revision metadata.
D:Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests. This examines model artifact image digest reproducibility rather than the specific mechanism reported here. It is not the decisive first check for model registry version provenance. The relevant evidence is: Review immutable artifact IDs, Model Registry versions, evaluation reports, approvals and deployed revision metadata.
Question 131
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A build pushes a model-serving image into production even after its dependency scan reports a critical vulnerability. The team requires an idempotent pipeline behavior and traceable evidence. Which TWO choices together provide the decisive diagnostic evidence and the most direct acceptance test? Select TWO.
- Inspect Cloud Build stage status propagation, vulnerability findings, policy thresholds and deployment triggers.
- A deliberately vulnerable test image is blocked from production while clean images pass the same gate.
- Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies.
- Training manifests exclude unlabeled recent transactions and measured metrics use verified ground truth.
- An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence.
Explanation
Correct Answers
A:Inspect Cloud Build stage status propagation, vulnerability findings, policy thresholds and deployment triggers. This inspects evidence relevant to cloud build container security gate. Security scanning provides protection only if the CI/CD pipeline enforces its result before release. After identifying the cause, the matching correction is: Fail protected build or release stages when prohibited vulnerabilities exist and require documented exceptions for approved deviations.
B:A deliberately vulnerable test image is blocked from production while clean images pass the same gate. This is specific acceptance evidence for cloud build container security gate. Security scanning provides protection only if the CI/CD pipeline enforces its result before release. It confirms the controlled correction: Fail protected build or release stages when prohibited vulnerabilities exist and require documented exceptions for approved deviations.
Incorrect Answers
C:Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. This relates to serving schema validation component but does not establish the requested before-and-after evidence. The actual diagnostic is: Inspect Cloud Build stage status propagation, vulnerability findings, policy thresholds and deployment triggers. The specific acceptance result is: A deliberately vulnerable test image is blocked from production while clean images pass the same gate.
D:Training manifests exclude unlabeled recent transactions and measured metrics use verified ground truth. This relates to scheduled retraining on delayed ground truth but does not establish the requested before-and-after evidence. The actual diagnostic is: Inspect Cloud Build stage status propagation, vulnerability findings, policy thresholds and deployment triggers. The specific acceptance result is: A deliberately vulnerable test image is blocked from production while clean images pass the same gate.
E:An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence. This relates to model evaluation gate before deployment but does not establish the requested before-and-after evidence. The actual diagnostic is: Inspect Cloud Build stage status propagation, vulnerability findings, policy thresholds and deployment triggers. The specific acceptance result is: A deliberately vulnerable test image is blocked from production while clean images pass the same gate.
Question 132
While reproducing a training orchestration incident in isolated staging, the following condition occurs: The same pipeline is manually edited between staging and production, so a critical feature filter differs across environments without a code review. The team requires an idempotent pipeline behavior and traceable evidence. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Inspect the pipeline DAG, dependency declarations, artifact references and execution ordering from recent runs.
- Train only on adequately matured labeled cohorts and monitor recent feature drift separately until outcomes arrive.
- Add a data-validation component that quarantines nonconforming inputs and fails downstream training until reviewed.
- Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments.
- Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
Explanation
Correct Answers
D:Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments. This inspects evidence relevant to pipeline environment parameter promotion. Promoting immutable pipeline definitions prevents untracked hand edits from changing processing semantics. After identifying the cause, the matching correction is: Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
E:Publish one versioned pipeline package and promote it with reviewed environment-specific parameters. This is the targeted corrective response for pipeline environment parameter promotion. Promoting immutable pipeline definitions prevents untracked hand edits from changing processing semantics. Verify the diagnosis using: Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments. Confirm success using: Staging and production runs report the same component digests and only documented configuration values differ.
Incorrect Answers
A:Inspect the pipeline DAG, dependency declarations, artifact references and execution ordering from recent runs. This is an action or diagnostic relevant to pipeline component dependency order, not one of the required two steps. In this scenario verify: Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments. Then correct: Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
B:Train only on adequately matured labeled cohorts and monitor recent feature drift separately until outcomes arrive. This is an action or diagnostic relevant to scheduled retraining on delayed ground truth, not one of the required two steps. In this scenario verify: Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments. Then correct: Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
C:Add a data-validation component that quarantines nonconforming inputs and fails downstream training until reviewed. This is an action or diagnostic relevant to data validation threshold gate, not one of the required two steps. In this scenario verify: Inspect component definitions, pipeline package versions, parameters and source revisions used in both environments. Then correct: Publish one versioned pipeline package and promote it with reviewed environment-specific parameters.
Question 133
During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A streaming feature pipeline aggregates transactions by processing arrival time, causing delayed events to land in the wrong customer-hour bucket. The handoff must preserve privacy, lineage and testable behavior. Which result is the strongest postchange evidence of success?
- A production model identifier resolves to its training execution, immutable artifacts, approvals and input dataset.
- For every row the feature version predates its prediction cutoff and prospective holdout quality is realistic.
- Annotators can complete validated labels without seeing identifying data and the project retains compliant traceability.
- Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
Explanation
Correct Answer
D:Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract. This is specific acceptance evidence for dataflow event-time window correctness. Arrival time and event time are different; event-time windows with lateness rules preserve business-time meaning. It confirms the controlled correction: Configure event-time windowing with an appropriate watermark and explicit late-data correction policy in the Dataflow pipeline.
Incorrect Answers
A:A production model identifier resolves to its training execution, immutable artifacts, approvals and input dataset. This result could demonstrate success for ml metadata lineage for production model but not for the stated dataflow event-time window correctness issue. The direct test is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
B:For every row the feature version predates its prediction cutoff and prospective holdout quality is realistic. This result could demonstrate success for point-in-time feature leakage prevention but not for the stated dataflow event-time window correctness issue. The direct test is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
C:Annotators can complete validated labels without seeing identifying data and the project retains compliant traceability. This result could demonstrate success for data minimization for labeling inputs but not for the stated dataflow event-time window correctness issue. The direct test is: Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
Question 134
During a regulated MLOps pipeline deployment, the following condition occurs: A freshly trained PyTorch model deploys but rejects the production client's JSON keys because only training unit tests ran in CI. The workflow must be repeatable and protect the last approved production model. Which configuration or procedure is the most appropriate correction?
- 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.
- Require inference contract and integration tests with representative requests before promoting each candidate artifact.
- Trigger candidate retraining only when justified evidence and sufficient labeled data are available, with evaluation gates.
Explanation
Correct Answer
C:Require inference contract and integration tests with representative requests before promoting each candidate artifact. This is the targeted corrective response for continuous delivery integration contract tests. Training success does not guarantee compatibility with the deployed model's consumer API. Verify the diagnosis using: Inspect request schema, container route behavior, test fixtures and CI stage dependency graph. Confirm success using: A deliberately incompatible payload prevents the release and a compatible fixture completes an end-to-end prediction.
Incorrect Answers
A:Make the load idempotent with keyed merge or overwrite-per-version semantics and safe retry handling. This action addresses airflow task idempotency on retry, not the failure in the described continuous delivery integration contract tests case. The targeted correction is: Require inference contract and integration tests with representative requests before promoting each candidate artifact.
B:Promote explicit immutable model and image versions with versioned release manifests and review gates. This action addresses cloud build release version promotion, not the failure in the described continuous delivery integration contract tests case. The targeted correction is: Require inference contract and integration tests with representative requests before promoting each candidate artifact.
D: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 continuous delivery integration contract tests case. The targeted correction is: Require inference contract and integration tests with representative requests before promoting each candidate artifact.
Question 135
At a Google Cloud analytics team preparing a governed pilot, the following condition occurs: An analyst normalizes numerical inputs in a notebook before training but later calls ML.PREDICT on raw table columns, shifting prediction values. The team must satisfy a reproducible low-code AI assessment. What implementation change resolves the specific issue without substituting an unrelated fix?
- Train a supervised BigQuery ML LOGISTIC_REG or supported classifier with the churn label and evaluate precision-recall behavior.
- Move deterministic feature transformations into the supported BigQuery ML TRANSFORM workflow and use the same model at inference.
- Use ML.PREDICT on the appropriate BigQuery ML model within a controlled SQL batch job, writing results to governed tables.
- Benchmark a supported lower-latency Gemini variant against the current model under representative traffic before routing requests.
Explanation
Correct Answer
B:Move deterministic feature transformations into the supported BigQuery ML TRANSFORM workflow and use the same model at inference. This is the targeted corrective response for bigquery ml transform serving parity. Transformations embedded with a BigQuery ML model reduce divergence between fitting and serving features. Verify the diagnosis using: Inspect the training preprocessing SQL, TRANSFORM clause, model input schema, and the inference query's feature values. Confirm success using: A controlled raw record yields the same transformed feature representation for evaluation and production SQL inference. Here, the relevant situation is: the following condition occurs: An analyst normalizes numerical inputs in a notebook before training but later calls….
Incorrect Answers
A:Train a supervised BigQuery ML LOGISTIC_REG or supported classifier with the churn label and evaluate precision-recall behavior. This action addresses bigquery ml logistic model for conversion probability, not the failure in the described bigquery ml transform serving parity case. The targeted correction is: Move deterministic feature transformations into the supported BigQuery ML TRANSFORM workflow and use the same model at inference.
C:Use ML.PREDICT on the appropriate BigQuery ML model within a controlled SQL batch job, writing results to governed tables. This action addresses bigquery ml sql batch prediction integration, not the failure in the described bigquery ml transform serving parity case. The targeted correction is: Move deterministic feature transformations into the supported BigQuery ML TRANSFORM workflow and use the same model at inference.
D:Benchmark a supported lower-latency Gemini variant against the current model under representative traffic before routing requests. This action addresses gemini model choice for latency versus complexity, not the failure in the described bigquery ml transform serving parity case. The targeted correction is: Move deterministic feature transformations into the supported BigQuery ML TRANSFORM workflow and use the same model at inference.
Question 136
During an independent acceptance rehearsal of a customer-facing AI capability, the following condition occurs: An accounts-payable team needs supplier names, invoice totals and line items but uses image classification to decide whether an invoice is present. The owners require measurable quality and controlled operating cost. Which underlying engineering principle explains why the original approach fails?
- A pretrained vision API may cover general image concepts, while specialized categories may need a custom model.
- A label-derived future field leaks the outcome into input features and makes offline metrics deceptively strong.
- Object or image classification only identifies image categories; document understanding targets structured text fields.
- Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary.
Explanation
Correct Answer
C:Object or image classification only identifies image categories; document understanding targets structured text fields. This explains the mechanism behind document ai structured invoice extraction and why the described failure is possible. A discriminating investigation is: Inspect representative invoices, desired structured fields, supported processor output and field-level extraction errors. The corresponding repair is: Use an appropriate Document AI processor for form or invoice extraction and validate required fields before downstream posting.
Incorrect Answers
A:A pretrained vision API may cover general image concepts, while specialized categories may need a custom model. This principle describes vision api pretrained image labels versus custom detection, which is not the cause of the observed document ai structured invoice extraction failure. The controlling explanation is: Object or image classification only identifies image categories; document understanding targets structured text fields.
B:A label-derived future field leaks the outcome into input features and makes offline metrics deceptively strong. This principle describes bigquery ml ml.evaluate leakage detection, which is not the cause of the observed document ai structured invoice extraction failure. The controlling explanation is: Object or image classification only identifies image categories; document understanding targets structured text fields.
D:Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary. This principle describes gemini model choice for latency versus complexity, which is not the cause of the observed document ai structured invoice extraction failure. The controlling explanation is: Object or image classification only identifies image categories; document understanding targets structured text fields.
Question 137
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A feature-building pipeline reuses a cached output after source data changed because the component's cache key does not incorporate dataset version. The team requires an idempotent pipeline behavior and traceable evidence. Select TWO: the decisive diagnostic step and the targeted repair for this specific issue.
- Require candidate evaluation against a governed baseline with an enforced acceptance threshold before promotion.
- Inspect component input parameters, artifact provenance, caching settings and source snapshot identifiers.
- Review schema tests, null-rate distributions, validation thresholds and which pipeline stages respond to failures.
- Promote explicit immutable model and image versions with versioned release manifests and review gates.
- Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
Explanation
Correct Answers
B:Inspect component input parameters, artifact provenance, caching settings and source snapshot identifiers. This inspects evidence relevant to kubeflow component cache invalidation. Pipeline caches assume identical declared inputs; hidden external dependencies can make cached outputs stale. After identifying the cause, the matching correction is: Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
E:Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching. This is the targeted corrective response for kubeflow component cache invalidation. Pipeline caches assume identical declared inputs; hidden external dependencies can make cached outputs stale. Verify the diagnosis using: Inspect component input parameters, artifact provenance, caching settings and source snapshot identifiers. Confirm success using: Changing the dataset version forces a rebuild while identical inputs may reuse validated cached results.
Incorrect Answers
A:Require candidate evaluation against a governed baseline with an enforced acceptance threshold before promotion. This is an action or diagnostic relevant to model evaluation gate before deployment, not one of the required two steps. In this scenario verify: Inspect component input parameters, artifact provenance, caching settings and source snapshot identifiers. Then correct: Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
C:Review schema tests, null-rate distributions, validation thresholds and which pipeline stages respond to failures. This is an action or diagnostic relevant to data validation threshold gate, not one of the required two steps. In this scenario verify: Inspect component input parameters, artifact provenance, caching settings and source snapshot identifiers. Then correct: Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
D: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 component input parameters, artifact provenance, caching settings and source snapshot identifiers. Then correct: Include immutable dataset versions and code revisions as explicit component inputs or disable inappropriate caching.
Question 138
During a staged restoration test at a secondary serving deployment, the following condition occurs: A new image classifier has only offline test results, yet engineers plan to replace every customer prediction immediately after deployment. The team must prove an actual user-visible improvement without weakening access controls. Which result is the strongest postchange evidence of success?
- Every controlled sample's returned label matches the model's trained class index with no misordered classes.
- Only a limited portion initially uses the candidate and a simulated regression triggers a tested rollback.
- Large documents finish without client timeout and consumers retrieve signed results using recorded job IDs.
- Latency and memory decline on target hardware while validation degradation stays within the approved threshold.
Explanation
Correct Answer
B:Only a limited portion initially uses the candidate and a simulated regression triggers a tested rollback. This is specific acceptance evidence for canary gradual traffic rollout. Progressive traffic exposure reduces blast radius compared with instant replacement of all production serving. It confirms the controlled correction: Roll out a small canary percentage first, monitor error and quality guardrails and expand only after acceptance.
Incorrect Answers
A:Every controlled sample's returned label matches the model's trained class index with no misordered classes. This result could demonstrate success for custom postprocessing label mapping drift but not for the stated canary gradual traffic rollout issue. The direct test is: Only a limited portion initially uses the candidate and a simulated regression triggers a tested rollback.
C:Large documents finish without client timeout and consumers retrieve signed results using recorded job IDs. This result could demonstrate success for online prediction timeout and batch fallback but not for the stated canary gradual traffic rollout issue. The direct test is: Only a limited portion initially uses the candidate and a simulated regression triggers a tested rollback.
D:Latency and memory decline on target hardware while validation degradation stays within the approved threshold. This result could demonstrate success for inference optimization using quantization but not for the stated canary gradual traffic rollout issue. The direct test is: Only a limited portion initially uses the candidate and a simulated regression triggers a tested rollback.
Question 139
During a production ML inference release on Google Cloud, the following condition occurs: Two releases use the same container tag latest, but the registry retargets it to a different dependency build between deployments. The service must meet its agreed quality and latency requirements. Which evidence most directly establishes what is wrong with this AI or ML workflow?
- Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
- Inspect startup logs, HTTP listening port, health-check expectations and request and response schema.
- Inspect traffic-splitting capabilities, release guardrails, rollback triggers and representative online metrics.
- Examine effective IAM bindings, invocation audit logs and the intended partner service account identity.
Explanation
Correct Answer
A:Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests. This inspects evidence relevant to model artifact image digest reproducibility. Mutable tags can point to new image content, defeating reproducibility even if the deployment text is unchanged. After identifying the cause, the matching correction is: Pin releases to reviewed immutable image digests and record image plus model artifact versions.
Incorrect Answers
B:Inspect startup logs, HTTP listening port, health-check expectations and request and response schema. This examines inference container health interface mismatch rather than the specific mechanism reported here. It is not the decisive first check for model artifact image digest reproducibility. The relevant evidence is: Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
C:Inspect traffic-splitting capabilities, release guardrails, rollback triggers and representative online metrics. This examines canary gradual traffic rollout rather than the specific mechanism reported here. It is not the decisive first check for model artifact image digest reproducibility. The relevant evidence is: Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
D:Examine effective IAM bindings, invocation audit logs and the intended partner service account identity. This examines public endpoint iam oversharing rather than the specific mechanism reported here. It is not the decisive first check for model artifact image digest reproducibility. The relevant evidence is: Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
Question 140
While preparing a production training workflow on Gemini Enterprise Agent Platform, the following condition occurs: An operations team uses a general-purpose chat model to forecast thousands of numeric hourly warehouse counts but ignores established daily and weekly seasonality. Engineers need an auditable, resource-aware correction to the issue. Which evidence most directly establishes what is wrong with this AI or ML workflow?
- Determine whether failures arise from changing reference knowledge, stable behavior patterns or inadequate retrieval coverage.
- Inspect time stamps, autocorrelation, horizon, seasonal cycles and out-of-sample forecast metrics against traditional baselines.
- Inspect output schema, decoding configuration, structured-output support and malformed samples under representative prompts.
- Compare pretraining benchmark scores, domain validation curves, trainable parameters and fine-tuning learning-rate choices.
Explanation
Correct Answer
B:Inspect time stamps, autocorrelation, horizon, seasonal cycles and out-of-sample forecast metrics against traditional baselines. This inspects evidence relevant to forecast trend and seasonality versus llm. Structured numeric forecasting relies on temporal evidence; a conversational model is not automatically the best forecaster. After identifying the cause, the matching correction is: Evaluate ARIMA_PLUS or an appropriate time-series forecasting model before adopting an expensive generative approach.
Incorrect Answers
A:Determine whether failures arise from changing reference knowledge, stable behavior patterns or inadequate retrieval coverage. This examines foundation model prompt versus fine-tuning rather than the specific mechanism reported here. It is not the decisive first check for forecast trend and seasonality versus llm. The relevant evidence is: Inspect time stamps, autocorrelation, horizon, seasonal cycles and out-of-sample forecast metrics against traditional baselines.
C:Inspect output schema, decoding configuration, structured-output support and malformed samples under representative prompts. This examines constrained generative json outputs rather than the specific mechanism reported here. It is not the decisive first check for forecast trend and seasonality versus llm. The relevant evidence is: Inspect time stamps, autocorrelation, horizon, seasonal cycles and out-of-sample forecast metrics against traditional baselines.
D:Compare pretraining benchmark scores, domain validation curves, trainable parameters and fine-tuning learning-rate choices. This examines fine-tuning catastrophic forgetting rather than the specific mechanism reported here. It is not the decisive first check for forecast trend and seasonality versus llm. The relevant evidence is: Inspect time stamps, autocorrelation, horizon, seasonal cycles and out-of-sample forecast metrics against traditional baselines.