Google Professional Machine Learning Engineer Practice Test 19
Question 180
During a regulated MLOps pipeline deployment, 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 workflow must be repeatable and protect the last approved production model. Which technical explanation best accounts for the reported behavior?
- Continuous training should not imply uncontrolled continuous deployment of every new model.
- Orchestration without failure notification can leave models stale even when the last successful deployment was correct.
- Model weights alone do not capture the data lineage necessary to reproduce and investigate predictions.
- Continuous training needs statistically and operationally justified policy thresholds rather than reactive promotion of noise.
Explanation
Correct Answer
B:Orchestration without failure notification can leave models stale even when the last successful deployment was correct. This explains the mechanism behind pipeline notification alert on silent failure and why the described failure is possible. A discriminating investigation is: Inspect task execution states, failed-job logs, alert routing, pipeline-age metrics and owner escalation settings. The corresponding repair is: Configure failure and stale-model alerts tied to actionable run-status and freshness service objectives.
Incorrect Answers
A:Continuous training should not imply uncontrolled continuous deployment of every new model. This principle describes continuous training rollback protection, which is not the cause of the observed pipeline notification alert on silent failure failure. The controlling explanation is: Orchestration without failure notification can leave models stale even when the last successful deployment was correct.
C:Model weights alone do not capture the data lineage necessary to reproduce and investigate predictions. This principle describes model feature lineage for audit, which is not the cause of the observed pipeline notification alert on silent failure failure. The controlling explanation is: Orchestration without failure notification can leave models stale even when the last successful deployment was correct.
D:Continuous training needs statistically and operationally justified policy thresholds rather than reactive promotion of noise. This principle describes canary failure feedback into ct policy, which is not the cause of the observed pipeline notification alert on silent failure failure. The controlling explanation is: Orchestration without failure notification can leave models stale even when the last successful deployment was correct.
Question 181
During a postdeployment AI safety and reliability review, the following condition occurs: A public conversational app relies on a single instruction asking the model to avoid prohibited content but does not test or screen generated responses. The model owner must provide repeatable evidence rather than a cosmetic dashboard change. Which result is the strongest postchange evidence of success?
- A controlled prompt-injection fixture is blocked or neutralized without disrupting ordinary allowed documents.
- A failing canary halts and restores the healthy model before broader traffic exposure.
- The candidate meets subgroup harm thresholds on representative held-out data without hiding global tradeoffs.
- Red-team test prompts trigger expected safeguards while legitimate requests continue to receive useful responses.
Explanation
Correct Answer
D:Red-team test prompts trigger expected safeguards while legitimate requests continue to receive useful responses. This is specific acceptance evidence for gen ai output safety policy enforcement. Prompt wording alone is not a reliable comprehensive safety boundary for public generative applications. It confirms the controlled correction: Apply supported safety filters and output policy checks with a controlled refusal or human review path for blocked results.
Incorrect Answers
A:A controlled prompt-injection fixture is blocked or neutralized without disrupting ordinary allowed documents. This result could demonstrate success for model armor untrusted content screening but not for the stated gen ai output safety policy enforcement issue. The direct test is: Red-team test prompts trigger expected safeguards while legitimate requests continue to receive useful responses.
B:A failing canary halts and restores the healthy model before broader traffic exposure. This result could demonstrate success for production incident rollback based on metrics but not for the stated gen ai output safety policy enforcement issue. The direct test is: Red-team test prompts trigger expected safeguards while legitimate requests continue to receive useful responses.
C:The candidate meets subgroup harm thresholds on representative held-out data without hiding global tradeoffs. This result could demonstrate success for model feature bias by subgroup but not for the stated gen ai output safety policy enforcement issue. The direct test is: Red-team test prompts trigger expected safeguards while legitimate requests continue to receive useful responses.
Question 182
During model-training validation on a Google Cloud experimental project, the following condition occurs: A retailer asks for next-week order value as a dollar amount, but the prototype returns yes/no purchase classes without estimated magnitude. The change should preserve a meaningful independent model-quality comparison. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Inspect required output type, business loss, sample labels, and the configured model objective and metric.
- Inspect task requirements for retrieval ranking, available query-document pairs, embedding similarity and labeled relevance examples.
- Reduce the learning rate and use justified parameter-efficient tuning or partial freezing with a controlled evaluation set.
- Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts.
- Compare an interpretable model or validated feature attribution approach with the black box and document decision tradeoffs.
Explanation
Correct Answers
A:Inspect required output type, business loss, sample labels, and the configured model objective and metric. This inspects evidence relevant to regression versus classification product choice. Categorical classification is not an estimator of an unrestricted continuous order value. After identifying the cause, the matching correction is: Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts.
D:Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts. This is the targeted corrective response for regression versus classification product choice. Categorical classification is not an estimator of an unrestricted continuous order value. Verify the diagnosis using: Inspect required output type, business loss, sample labels, and the configured model objective and metric. Confirm success using: Held-out predictions are currency-valued and their absolute or squared errors match the financial evaluation contract.
Incorrect Answers
B:Inspect task requirements for retrieval ranking, available query-document pairs, embedding similarity and labeled relevance examples. This is an action or diagnostic relevant to embedding retrieval versus text classifier, not one of the required two steps. In this scenario verify: Inspect required output type, business loss, sample labels, and the configured model objective and metric. Then correct: Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts.
C:Reduce the learning rate and use justified parameter-efficient tuning or partial freezing with a controlled evaluation set. This is an action or diagnostic relevant to fine-tuning catastrophic forgetting, not one of the required two steps. In this scenario verify: Inspect required output type, business loss, sample labels, and the configured model objective and metric. Then correct: Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts.
E:Compare an interpretable model or validated feature attribution approach with the black box and document decision tradeoffs. This is an action or diagnostic relevant to interpretability-constrained model baseline, not one of the required two steps. In this scenario verify: Inspect required output type, business loss, sample labels, and the configured model objective and metric. Then correct: Select a regression architecture and matching numeric loss rather than using a binary classifier for amounts.
Question 183
During a staged restoration test at a secondary serving deployment, 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 team must prove an actual user-visible improvement without weakening access controls. What should the engineer inspect before selecting a correction?
- Compare Artifact Registry image digest, build provenance, runtime dependency inventory and release manifests.
- Inspect experiment assignment IDs, routing hashes, variant weights and per-user exposure histories.
- Inspect resolved endpoint address, authorized network routing, service access policy and connection logs.
- Compare tokenizer hash, normalization rules, token IDs and inference schema against the trained model artifacts.
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. This matters here because during a staged restoration test at a secondary serving deployment, the following condition occurs: Two releases use….
Incorrect Answers
B:Inspect experiment assignment IDs, routing hashes, variant weights and per-user exposure histories. This examines a/b assignment consistency for user journeys 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 resolved endpoint address, authorized network routing, service access policy and connection logs. This examines private vpc routing failure for serving 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:Compare tokenizer hash, normalization rules, token IDs and inference schema against the trained model artifacts. This examines inference preprocessing tokenization 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.
Question 184
During an independent acceptance rehearsal of a customer-facing AI capability, the following condition occurs: A commerce analyst uses a binary classifier to predict exact revenue per order even though labels are continuous currency amounts. The owners require measurable quality and controlled operating cost. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata.
- Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.
- Use ML.PREDICT on the appropriate BigQuery ML model within a controlled SQL batch job, writing results to governed tables.
- Inspect the training preprocessing SQL, TRANSFORM clause, model input schema, and the inference query's feature values.
- Index versioned policy sources and use retrieval-grounded Gemini responses for changing facts, reserving tuning for justified behavior adaptation.
Explanation
Correct Answers
A:Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata. This inspects evidence relevant to bigquery ml linear regression for continuous value. Classification labels cannot directly represent unrestricted continuous outcomes; SQL regression models match numeric targets. After identifying the cause, the matching correction is: Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.
B:Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records. This is the targeted corrective response for bigquery ml linear regression for continuous value. Classification labels cannot directly represent unrestricted continuous outcomes; SQL regression models match numeric targets. Verify the diagnosis using: Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata. Confirm success using: Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values.
Incorrect Answers
C:Use ML.PREDICT on the appropriate BigQuery ML model within a controlled SQL batch job, writing results to governed tables. This is an action or diagnostic relevant to bigquery ml sql batch prediction integration, not one of the required two steps. In this scenario verify: Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata. Then correct: Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.
D:Inspect the training preprocessing SQL, TRANSFORM clause, model input schema, and the inference query's feature values. This is an action or diagnostic relevant to bigquery ml transform serving parity, not one of the required two steps. In this scenario verify: Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata. Then correct: Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.
E:Index versioned policy sources and use retrieval-grounded Gemini responses for changing facts, reserving tuning for justified behavior adaptation. This is an action or diagnostic relevant to foundation model tuning versus retrieval update, not one of the required two steps. In this scenario verify: Inspect the target column type, business loss metric, and sample prediction schema alongside model_type in the BigQuery model metadata. Then correct: Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.
Question 185
During a staged restoration test at a secondary serving deployment, the following condition occurs: A text classifier was trained with a specific subword vocabulary, but production preprocessors use a newer incompatible tokenizer. The team must prove an actual user-visible improvement without weakening access controls. Which configuration or procedure is the most appropriate correction?
- Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests.
- 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.
- Deploy a supported private endpoint or constrained private network path with least-privilege caller authorization.
Explanation
Correct Answer
A:Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests. This is the targeted corrective response for inference preprocessing tokenization mismatch. Embedding matrices interpret token IDs according to one vocabulary; mismatched preprocessing changes input meaning. Verify the diagnosis using: Compare tokenizer hash, normalization rules, token IDs and inference schema against the trained model artifacts. Confirm success using: Identical sample text yields matching token IDs and model outputs in offline and online environments.
Incorrect Answers
B:Benchmark a Cloud Run custom container against a managed Agent Platform endpoint and choose for workload fit. This action addresses cloud run versus managed inference endpoint, not the failure in the described inference preprocessing tokenization mismatch case. The targeted correction is: Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests.
C:Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency. This action addresses dynamic batch serving for gpu saturation, not the failure in the described inference preprocessing tokenization mismatch case. The targeted correction is: Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests.
D:Deploy a supported private endpoint or constrained private network path with least-privilege caller authorization. This action addresses private endpoint for regulated inference traffic, not the failure in the described inference preprocessing tokenization mismatch case. The targeted correction is: Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests.
Question 186
During a staged restoration test at a secondary serving deployment, the following condition occurs: A model server scales down to zero expensive workers overnight but cannot meet the first morning burst because loading weights takes several minutes. The team must prove an actual user-visible improvement without weakening access controls. Which outcome most directly demonstrates a successful, measurable fix?
- The camera processes representative frames inside its latency budget during an intentional network outage.
- Previously missed customer IDs retrieve the intended feature values without collapsing distinct leading-zero identifiers.
- The morning surge meets availability and p95 latency objectives while off-peak cost remains documented.
- Latency and memory decline on target hardware while validation degradation stays within the approved threshold.
Explanation
Correct Answer
C:The morning surge meets availability and p95 latency objectives while off-peak cost remains documented. This is specific acceptance evidence for autoscaling minimum capacity versus cold load. Reactive scaling cannot instantly serve requests when cold model initialization is long. It confirms the controlled correction: Set a justified warm minimum or scheduled capacity and balance initialization overhead against idle cost. This matters here because during a staged restoration test at a secondary serving deployment, the following condition occurs: A model server….
Incorrect Answers
A:The camera processes representative frames inside its latency budget during an intentional network outage. This result could demonstrate success for edge versus cloud inference network constraints but not for the stated autoscaling minimum capacity versus cold load issue. The direct test is: The morning surge meets availability and p95 latency objectives while off-peak cost remains documented.
B:Previously missed customer IDs retrieve the intended feature values without collapsing distinct leading-zero identifiers. This result could demonstrate success for feature online entity key mismatch but not for the stated autoscaling minimum capacity versus cold load issue. The direct test is: The morning surge meets availability and p95 latency objectives while off-peak cost remains documented. This matters here because during a staged restoration test at a secondary serving deployment, the following condition occurs: A model server….
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 autoscaling minimum capacity versus cold load issue. The direct test is: The morning surge meets availability and p95 latency objectives while off-peak cost remains documented.
Question 187
During a regulated MLOps pipeline deployment, the following condition occurs: A CI/CD workflow always deploys the most recent checkpoint even when its validation F1 is below the previous approved baseline. The workflow must be repeatable and protect the last approved production model. Which result is the strongest postchange evidence of success?
- A controlled worker loss resumes from saved progress and produces consistent output counts.
- An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence.
- No-op periods create no unnecessary models and validated drift events produce candidates requiring approval.
- An incompatible payload fails visibly while compatible older versions follow documented conversion rules.
Explanation
Correct Answer
B:An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence. This is specific acceptance evidence for model evaluation gate before deployment. Evaluation without a blocking release decision cannot protect production from quality regressions. It confirms the controlled correction: Require candidate evaluation against a governed baseline with an enforced acceptance threshold before promotion.
Incorrect Answers
A:A controlled worker loss resumes from saved progress and produces consistent output counts. This result could demonstrate success for ray distributed task failure isolation but not for the stated model evaluation gate before deployment issue. The direct test is: An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence.
C:No-op periods create no unnecessary models and validated drift events produce candidates requiring approval. This result could demonstrate success for data drift triggered retraining policy but not for the stated model evaluation gate before deployment issue. The direct test is: An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence.
D:An incompatible payload fails visibly while compatible older versions follow documented conversion rules. This result could demonstrate success for serving schema validation component but not for the stated model evaluation gate before deployment issue. The direct test is: An intentionally weak candidate cannot deploy, whereas an approved passing model can proceed with traceable evidence.
Question 188
During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A numeric risk feature contains occasional physically impossible sensor spikes that inflate its fitted scale and degrade otherwise normal predictions. The handoff must preserve privacy, lineage and testable behavior. Which technical explanation best accounts for the reported behavior?
- Long-lived secrets copied into notebooks or Git can be reused outside intended workloads and must be rotated.
- A scaler learned from corrupted or held-out values can distort typical feature magnitudes and leak evaluation information.
- Skew concentrates otherwise distributed work on a single hot key, which extra workers alone may not resolve.
- Collaborative notebooks execute with real identities; editor convenience does not justify unrestricted production-data access.
Explanation
Correct Answer
B:A scaler learned from corrupted or held-out values can distort typical feature magnitudes and leak evaluation information. This explains the mechanism behind outlier handling before feature standardization and why the described failure is possible. A discriminating investigation is: Inspect feature quantiles, physical validity bounds, missingness and training-only fit statistics against independent validation data. The corresponding repair is: Quarantine invalid spikes and fit a robust transformation using training data only, then apply it consistently at inference.
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 outlier handling before feature standardization failure. The controlling explanation is: A scaler learned from corrupted or held-out values can distort typical feature magnitudes and leak evaluation information.
C:Skew concentrates otherwise distributed work on a single hot key, which extra workers alone may not resolve. This principle describes managed spark join-key skew repair, which is not the cause of the observed outlier handling before feature standardization failure. The controlling explanation is: A scaler learned from corrupted or held-out values can distort typical feature magnitudes and leak evaluation information.
D:Collaborative notebooks execute with real identities; editor convenience does not justify unrestricted production-data access. This principle describes colab enterprise notebook data access scope, which is not the cause of the observed outlier handling before feature standardization failure. The controlling explanation is: A scaler learned from corrupted or held-out values can distort typical feature magnitudes and leak evaluation information.
Question 189
During model-training validation on a Google Cloud experimental project, 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. The change should preserve a meaningful independent model-quality comparison. Which outcome most directly demonstrates a successful, measurable fix?
- Every evaluation feature is available at its simulated prediction time and later-period scores remain stable.
- Relevant articles rank highly for unseen paraphrased queries and retrieval hit rate is measured against judgments.
- The selected candidate meets fraud recall, inference latency and cost limits on a representative untouched split.
- Rolling-origin backtests improve horizon error and seasonal intervals without inventing unsupported demand figures.
Explanation
Correct Answer
D:Rolling-origin backtests improve horizon error and seasonal intervals without inventing unsupported demand figures. This is specific acceptance evidence for forecast trend and seasonality versus llm. Structured numeric forecasting relies on temporal evidence; a conversational model is not automatically the best forecaster. It confirms the controlled correction: Evaluate ARIMA_PLUS or an appropriate time-series forecasting model before adopting an expensive generative approach.
Incorrect Answers
A:Every evaluation feature is available at its simulated prediction time and later-period scores remain stable. This result could demonstrate success for temporal training label leakage in rolling windows but not for the stated forecast trend and seasonality versus llm issue. The direct test is: Rolling-origin backtests improve horizon error and seasonal intervals without inventing unsupported demand figures.
B:Relevant articles rank highly for unseen paraphrased queries and retrieval hit rate is measured against judgments. This result could demonstrate success for embedding retrieval versus text classifier but not for the stated forecast trend and seasonality versus llm issue. The direct test is: Rolling-origin backtests improve horizon error and seasonal intervals without inventing unsupported demand figures.
C:The selected candidate meets fraud recall, inference latency and cost limits on a representative untouched split. This result could demonstrate success for deep model versus gradient boosted trees on tabular but not for the stated forecast trend and seasonality versus llm issue. The direct test is: Rolling-origin backtests improve horizon error and seasonal intervals without inventing unsupported demand figures.
Question 190
During model-training validation on a Google Cloud experimental project, the following condition occurs: A model fits in one GPU but a huge dataset takes too long to process, and engineers propose splitting its layers between devices without memory pressure. The change should preserve a meaningful independent model-quality comparison. Which TWO selections represent the correct investigation and remediation together? Select TWO.
- Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators.
- Deduplicate or correctly weight validation samples when aggregating distributed metrics across workers.
- Inspect gradient ranges, loss-scaling logs, overflow events and the affected operations with test batches.
- Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.
- Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck.
Explanation
Correct Answers
A:Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators. This inspects evidence relevant to distributed data parallel gradient synchronization. When a model fits on each worker, partitioning training data is often more suitable than splitting the model itself. After identifying the cause, the matching correction is: Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.
D:Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency. This is the targeted corrective response for distributed data parallel gradient synchronization. When a model fits on each worker, partitioning training data is often more suitable than splitting the model itself. Verify the diagnosis using: Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators. Confirm success using: More replicas shorten epoch time without materially changing convergence or causing network communication to dominate.
Incorrect Answers
B:Deduplicate or correctly weight validation samples when aggregating distributed metrics across workers. This is an action or diagnostic relevant to data-parallel evaluation duplicate samples, not one of the required two steps. In this scenario verify: Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators. Then correct: Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.
C:Inspect gradient ranges, loss-scaling logs, overflow events and the affected operations with test batches. This is an action or diagnostic relevant to mixed precision overflow diagnosis, not one of the required two steps. In this scenario verify: Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators. Then correct: Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.
E:Tune supported distributed batch sizing, communication strategy or placement after measuring the bottleneck. This is an action or diagnostic relevant to distributed gradient communication bottleneck, not one of the required two steps. In this scenario verify: Measure model footprint, per-device utilization, communication cost and batch throughput on one and multiple accelerators. Then correct: Prefer suitable data parallel replicas with synchronized gradient aggregation, then benchmark scale efficiency.