Google Professional Machine Learning Engineer Practice Test 93
Question 920
As the team tests a new governed AI release against adversarial and normal data, the following condition occurs: A production evaluator cannot match delayed loan repayment outcomes to the predictions that were made weeks earlier because requests lack stable IDs. The fix must retain legitimate model usefulness and appropriate controls. Which targeted engineering change most directly addresses the failure?
- Version a single feature contract and add automated parity tests to the preprocessing pipeline and serving workflow.
- Require validated business quality and technical reliability thresholds to continue rollout, with tested rollback on failure.
- Capture immutable prediction identifiers and join verified outcome records to the corresponding earlier inference request.
- Evaluate and address documented subgroup disparities using appropriate fairness metrics and approved data/model changes.
Explanation
Correct Answer
C:Capture immutable prediction identifiers and join verified outcome records to the corresponding earlier inference request. This is the targeted corrective response for ground truth association for quality monitoring. Supervised production quality requires reliable prediction-to-ground-truth linkage, not only unlabeled model outputs. Verify the diagnosis using: Inspect inference event IDs, timestamps, later label keys and the quality monitoring join schema. Confirm success using: An evaluation fixture correctly associates each matured label with exactly one prior prediction. This matters here because as the team tests a new governed AI release against adversarial and normal data, the following condition occurs: A….
Incorrect Answers
A:Version a single feature contract and add automated parity tests to the preprocessing pipeline and serving workflow. This action addresses training-serving feature skew detection, not the failure in the described ground truth association for quality monitoring case. The targeted correction is: Capture immutable prediction identifiers and join verified outcome records to the corresponding earlier inference request.
B:Require validated business quality and technical reliability thresholds to continue rollout, with tested rollback on failure. This action addresses production incident rollback based on metrics, not the failure in the described ground truth association for quality monitoring case. The targeted correction is: Capture immutable prediction identifiers and join verified outcome records to the corresponding earlier inference request. This matters here because as the team tests a new governed AI release against adversarial and normal data, the following condition occurs: A….
D:Evaluate and address documented subgroup disparities using appropriate fairness metrics and approved data/model changes. This action addresses model feature bias by subgroup, not the failure in the described ground truth association for quality monitoring case. The targeted correction is: Capture immutable prediction identifiers and join verified outcome records to the corresponding earlier inference request.
Question 921
During a regulated MLOps pipeline deployment, 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 workflow must be repeatable and protect the last approved production model. What acceptance result verifies the intended behavior after remediation?
- A controlled worker loss resumes from saved progress and produces consistent output counts.
- A deliberately malformed dataset is rejected before training while healthy fixtures proceed successfully.
- Small harmless fluctuations do not force releases and genuine regressions invoke a tested rollback or controlled retraining.
- Staging and production runs report the same component digests and only documented configuration values differ.
Explanation
Correct Answer
D:Staging and production runs report the same component digests and only documented configuration values differ. This is specific acceptance evidence for pipeline environment parameter promotion. Promoting immutable pipeline definitions prevents untracked hand edits from changing processing semantics. It confirms the controlled correction: Publish one versioned pipeline package and promote it with reviewed environment-specific parameters. This matters here because the same pipeline is manually edited between staging and production, so a critical feature filter differs across….
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 pipeline environment parameter promotion issue. The direct test is: Staging and production runs report the same component digests and only documented configuration values differ.
B:A deliberately malformed dataset is rejected before training while healthy fixtures proceed successfully. This result could demonstrate success for data validation threshold gate but not for the stated pipeline environment parameter promotion issue. The direct test is: Staging and production runs report the same component digests and only documented configuration values differ.
C:Small harmless fluctuations do not force releases and genuine regressions invoke a tested rollback or controlled retraining. This result could demonstrate success for canary failure feedback into ct policy but not for the stated pipeline environment parameter promotion issue. The direct test is: Staging and production runs report the same component digests and only documented configuration values differ.
Question 922
While reproducing a training orchestration incident in isolated staging, the following condition occurs: A data ingestion pipeline changes a numeric feature into a string, but the model-serving path accepts it and silently substitutes missing values. The team requires an idempotent pipeline behavior and traceable evidence. Which TWO complementary choices jointly establish the evidence and appropriate correction? Select TWO.
- Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
- Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact.
- Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies.
- Inspect the pipeline DAG, dependency declarations, artifact references and execution ordering from recent runs.
- Retain the last healthy model and require canary validation before switching all prediction traffic.
Explanation
Correct Answers
A:Enforce a shared typed schema contract and reject incompatible feature versions before training or inference. This is the targeted corrective response for serving schema validation component. Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries. Verify the diagnosis using: Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. Confirm success using: An incompatible payload fails visibly while compatible older versions follow documented conversion rules.
C:Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. This inspects evidence relevant to serving schema validation component. Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries. After identifying the cause, the matching correction is: Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
Incorrect Answers
B:Record governed dataset snapshot and transformation versions as pipeline inputs linked to the resulting model artifact. This is an action or diagnostic relevant to model feature lineage for audit, not one of the required two steps. In this scenario verify: Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. Then correct: Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
D: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 incoming schema versions, validation logs, casting rules and downstream prediction anomalies. Then correct: Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
E:Retain the last healthy model and require canary validation before switching all prediction traffic. This is an action or diagnostic relevant to continuous training rollback protection, not one of the required two steps. In this scenario verify: Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. Then correct: Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.
Question 923
During a production ML inference release on Google Cloud, the following condition occurs: An industrial camera must classify frames while offline for hours, but its prototype requires every frame to reach a remote cloud endpoint. The service must meet its agreed quality and latency requirements. Which diagnostic check should be performed FIRST to isolate the described cause?
- Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility.
- Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.
- Inspect quota errors, retry behavior, queue occupancy, tail latency and autoscaling upper limits.
- Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.
Explanation
Correct Answer
A:Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility. This inspects evidence relevant to edge versus cloud inference network constraints. A cloud-only inference architecture cannot satisfy hard offline and low-latency edge requirements. After identifying the cause, the matching correction is: Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns. This matters here because an industrial camera must classify frames while offline for hours, but its prototype requires every frame to reach a….
Incorrect Answers
B:Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract. This examines cloud run versus managed inference endpoint rather than the specific mechanism reported here. It is not the decisive first check for edge versus cloud inference network constraints. The relevant evidence is: Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility.
C:Inspect quota errors, retry behavior, queue occupancy, tail latency and autoscaling upper limits. This examines predictive serving rate limit and backpressure rather than the specific mechanism reported here. It is not the decisive first check for edge versus cloud inference network constraints. The relevant evidence is: Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility.
D:Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements. This examines private endpoint for regulated inference traffic rather than the specific mechanism reported here. It is not the decisive first check for edge versus cloud inference network constraints. The relevant evidence is: Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility.
Question 924
During model-training validation on a Google Cloud experimental project, the following condition occurs: A transformer cannot fit its weights and activations on one device even at minimal batch size, but the team proposes more data replicas only. The change should preserve a meaningful independent model-quality comparison. Which result is the strongest postchange evidence of success?
- The chosen accelerator meets throughput and convergence expectations without unsupported operations or unstable retracing.
- Domain accuracy improves while broad language capability remains within the agreed retention threshold.
- Scale efficiency improves and model quality remains stable without synchronization stalls dominating every step.
- The model fits and trains across devices with controlled synchronization overhead and reproducible checkpoints.
Explanation
Correct Answer
D:The model fits and trains across devices with controlled synchronization overhead and reproducible checkpoints. This is specific acceptance evidence for model parallelism for memory-bound transformer. Data parallel replicas each hold a model copy; they do not solve a model-too-large-for-one-device problem. It confirms the controlled correction: Evaluate supported tensor or pipeline model parallelism to split the model across compatible accelerators. This matters here because during model-training validation on a Google Cloud experimental project, the following condition occurs: A….
Incorrect Answers
A:The chosen accelerator meets throughput and convergence expectations without unsupported operations or unstable retracing. This result could demonstrate success for tpu versus gpu for tensorflow workload but not for the stated model parallelism for memory-bound transformer issue. The direct test is: The model fits and trains across devices with controlled synchronization overhead and reproducible checkpoints.
B:Domain accuracy improves while broad language capability remains within the agreed retention threshold. This result could demonstrate success for fine-tuning catastrophic forgetting but not for the stated model parallelism for memory-bound transformer issue. The direct test is: The model fits and trains across devices with controlled synchronization overhead and reproducible checkpoints.
C:Scale efficiency improves and model quality remains stable without synchronization stalls dominating every step. This result could demonstrate success for distributed gradient communication bottleneck but not for the stated model parallelism for memory-bound transformer issue. The direct test is: The model fits and trains across devices with controlled synchronization overhead and reproducible checkpoints.
Question 925
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 outcome most directly demonstrates a successful, measurable fix?
- The notebook still trains on approved sample tables while restricted production datasets deny unauthorized reads.
- Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation.
- A production model identifier resolves to its training execution, immutable artifacts, approvals and input dataset.
- Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract.
Explanation
Correct Answer
B:Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation. This is specific acceptance evidence for managed spark join-key skew repair. Skew concentrates otherwise distributed work on a single hot key, which extra workers alone may not resolve. It confirms the controlled correction: Handle the dominant key separately or repartition and salt the skewed join while validating join semantics and counts. This matters here because within a jointly maintained Google Cloud data science workspace, the following condition occurs: A distributed….
Incorrect Answers
A:The notebook still trains on approved sample tables while restricted production datasets deny unauthorized reads. This result could demonstrate success for colab enterprise notebook data access scope but not for the stated managed spark join-key skew repair issue. The direct test is: Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation. This matters here because within a jointly maintained Google Cloud data science workspace, the following condition occurs: A distributed….
C: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 managed spark join-key skew repair issue. The direct test is: Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation.
D:Out-of-order sample events produce the intended hourly totals and late-data handling follows the documented contract. This result could demonstrate success for dataflow event-time window correctness but not for the stated managed spark join-key skew repair issue. The direct test is: Long-tail shuffle tasks shorten and the normalized output row count matches a reference implementation.
Question 926
During a staged restoration test at a secondary serving deployment, the following condition occurs: An industrial camera must classify frames while offline for hours, but its prototype requires every frame to reach a remote cloud endpoint. The team must prove an actual user-visible improvement without weakening access controls. Which configuration or procedure is the most appropriate correction?
- Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns.
- Restrict invoke access to the necessary principal and endpoint, using appropriate short-lived authentication.
- Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests.
- Promote an explicitly approved model version from Agent Platform Model Registry and record deployment lineage.
Explanation
Correct Answer
A:Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns. This is the targeted corrective response for edge versus cloud inference network constraints. A cloud-only inference architecture cannot satisfy hard offline and low-latency edge requirements. Verify the diagnosis using: Measure bandwidth, network availability, latency deadline, edge hardware limits and model compression compatibility. Confirm success using: The camera processes representative frames inside its latency budget during an intentional network outage.
Incorrect Answers
B:Restrict invoke access to the necessary principal and endpoint, using appropriate short-lived authentication. This action addresses public endpoint iam oversharing, not the failure in the described edge versus cloud inference network constraints case. The targeted correction is: Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns.
C:Package and version the exact trained preprocessing pipeline with the served model and run equivalence tests. This action addresses inference preprocessing tokenization mismatch, not the failure in the described edge versus cloud inference network constraints case. The targeted correction is: Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns.
D:Promote an explicitly approved model version from Agent Platform Model Registry and record deployment lineage. This action addresses model registry version provenance, not the failure in the described edge versus cloud inference network constraints case. The targeted correction is: Deploy a verified edge-compatible model for required offline decisions and sync summarized results when connectivity returns.
Question 927
During a staged restoration test at a secondary serving deployment, the following condition occurs: A GPU inference server shows low CPU use but increasing queue length and slow responses under request bursts; the only scaling rule watches CPU. The team must prove an actual user-visible improvement without weakening access controls. Which TWO choices together provide the decisive diagnostic evidence and the most direct acceptance test? Select TWO.
- A known prediction request succeeds through the deployed endpoint and readiness reflects successful initialization.
- Peak requests trigger additional serving capacity and tail latency recovers without repeated scale oscillations.
- Compare request rate, queue depth, GPU utilization, per-instance concurrency and p95 response latency.
- Previously missed customer IDs retrieve the intended feature values without collapsing distinct leading-zero identifiers.
- Examine effective IAM bindings, invocation audit logs and the intended partner service account identity.
Explanation
Correct Answers
B:Peak requests trigger additional serving capacity and tail latency recovers without repeated scale oscillations. This is specific acceptance evidence for throughput-based autoscaling versus cpu metric. CPU usage can be weakly related to accelerator or queue-bound inference workloads. It confirms the controlled correction: Configure an appropriate request-concurrency or workload-driven scaling policy with resource and cooldown limits.
C:Compare request rate, queue depth, GPU utilization, per-instance concurrency and p95 response latency. This inspects evidence relevant to throughput-based autoscaling versus cpu metric. CPU usage can be weakly related to accelerator or queue-bound inference workloads. After identifying the cause, the matching correction is: Configure an appropriate request-concurrency or workload-driven scaling policy with resource and cooldown limits.
Incorrect Answers
A:A known prediction request succeeds through the deployed endpoint and readiness reflects successful initialization. This relates to inference container health interface mismatch but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare request rate, queue depth, GPU utilization, per-instance concurrency and p95 response latency. The specific acceptance result is: Peak requests trigger additional serving capacity and tail latency recovers without repeated scale oscillations.
D:Previously missed customer IDs retrieve the intended feature values without collapsing distinct leading-zero identifiers. This relates to feature online entity key mismatch but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare request rate, queue depth, GPU utilization, per-instance concurrency and p95 response latency. The specific acceptance result is: Peak requests trigger additional serving capacity and tail latency recovers without repeated scale oscillations.
E:Examine effective IAM bindings, invocation audit logs and the intended partner service account identity. This relates to public endpoint iam oversharing but does not establish the requested before-and-after evidence. The actual diagnostic is: Compare request rate, queue depth, GPU utilization, per-instance concurrency and p95 response latency. The specific acceptance result is: Peak requests trigger additional serving capacity and tail latency recovers without repeated scale oscillations.
Question 928
During a postdeployment AI safety and reliability review, the following condition occurs: A company assistant follows instructions embedded in an untrusted retrieved webpage, overriding the system-approved task and attempting to reveal internal data. The model owner must provide repeatable evidence rather than a cosmetic dashboard change. Which TWO complementary choices jointly establish the evidence and appropriate correction? Select TWO.
- Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.
- Restrict dataset and bucket access by principal and resource; enforce approved egress controls and audit sensitive reads.
- Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening.
- Inspect attachment ingestion, trust boundaries, malicious prompt examples and deployed Model Armor filter results.
- Require validated business quality and technical reliability thresholds to continue rollout, with tested rollback on failure.
Explanation
Correct Answers
A:Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. This inspects evidence relevant to prompt injection through retrieved documents. External retrieval content can carry attacker instructions; data should not receive the authority of trusted application instructions. After identifying the cause, the matching correction is: Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening. This matters here because a company assistant follows instructions embedded in an untrusted retrieved webpage, overriding the system-approved….
C:Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening. This is the targeted corrective response for prompt injection through retrieved documents. External retrieval content can carry attacker instructions; data should not receive the authority of trusted application instructions. Verify the diagnosis using: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. Confirm success using: Attack-bearing documents fail to redirect privileged behavior while legitimate grounded answers remain available. This matters here because a company assistant follows instructions embedded in an untrusted retrieved webpage, overriding the system-approved….
Incorrect Answers
B:Restrict dataset and bucket access by principal and resource; enforce approved egress controls and audit sensitive reads. This is an action or diagnostic relevant to training data exfiltration through broad access, not one of the required two steps. In this scenario verify: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. Then correct: Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening. This matters here because a company assistant follows instructions embedded in an untrusted retrieved webpage, overriding the system-approved….
D:Inspect attachment ingestion, trust boundaries, malicious prompt examples and deployed Model Armor filter results. This is an action or diagnostic relevant to model armor untrusted content screening, not one of the required two steps. In this scenario verify: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. Then correct: Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening.
E:Require validated business quality and technical reliability thresholds to continue rollout, with tested rollback on failure. This is an action or diagnostic relevant to production incident rollback based on metrics, not one of the required two steps. In this scenario verify: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions. Then correct: Treat retrieved text as untrusted data, isolate instructions, restrict tool actions and apply appropriate Model Armor screening.
Question 929
At a Google Cloud analytics team preparing a governed pilot, the following condition occurs: A high-volume classification API uses a slow reasoning-optimized foundation model for simple short labels even though a smaller supported model may suffice. The team must satisfy a reproducible low-code AI assessment. Which statement identifies the primary reason for this failure?
- Unfocused context consumes tokens and can distract generation; targeted retrieval can reduce cost without discarding needed facts.
- Clustering segments observations without learning their known outcomes, while supervised classification predicts those outcomes.
- Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary.
- Object or image classification only identifies image categories; document understanding targets structured text fields.
Explanation
Correct Answer
C:Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary. This explains the mechanism behind gemini model choice for latency versus complexity and why the described failure is possible. A discriminating investigation is: Inspect class definitions, sample error costs, throughput, p95 latency, pricing and published model capability constraints. The corresponding repair is: Benchmark a supported lower-latency Gemini variant against the current model under representative traffic before routing requests.
Incorrect Answers
A:Unfocused context consumes tokens and can distract generation; targeted retrieval can reduce cost without discarding needed facts. This principle describes gemini prompt context token budget, which is not the cause of the observed gemini model choice for latency versus complexity failure. The controlling explanation is: Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary.
B:Clustering segments observations without learning their known outcomes, while supervised classification predicts those outcomes. This principle describes bigquery ml logistic model for conversion probability, which is not the cause of the observed gemini model choice for latency versus complexity failure. The controlling explanation is: Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary.
D:Object or image classification only identifies image categories; document understanding targets structured text fields. This principle describes document ai structured invoice extraction, which is not the cause of the observed gemini model choice for latency versus complexity failure. The controlling explanation is: Production model selection is a task-quality and cost tradeoff; using the largest model by default is unnecessary.
Question 930
While preparing a production training workflow on Gemini Enterprise Agent Platform, the following condition occurs: A team wants to tune an image-generation foundation model using scraped brand imagery without verifying usage rights or consent. Engineers need an auditable, resource-aware correction to the issue. What acceptance result verifies the intended behavior after remediation?
- The training manifest contains traceable permitted sources and safety tests detect known prohibited outputs.
- The job schedules successfully with the approved hardware and remains inside budget and residency constraints.
- No duplicated or synthetic neighbor crosses the evaluation boundary and rare-class recall reflects unseen defects.
- More replicas shorten epoch time without materially changing convergence or causing network communication to dominate.
Explanation
Correct Answer
A:The training manifest contains traceable permitted sources and safety tests detect known prohibited outputs. This is specific acceptance evidence for fine-tuning image generator data rights. Fine-tuning technical feasibility does not confer rights to reuse restricted training images or ignore model safety. It confirms the controlled correction: Use licensed and permitted training images with appropriate governance reviews before any supported tuning job. This matters here because while preparing a production training workflow on Gemini Enterprise Agent Platform, the following condition occurs:….
Incorrect Answers
B:The job schedules successfully with the approved hardware and remains inside budget and residency constraints. This result could demonstrate success for accelerator quota and regional availability but not for the stated fine-tuning image generator data rights issue. The direct test is: The training manifest contains traceable permitted sources and safety tests detect known prohibited outputs.
C:No duplicated or synthetic neighbor crosses the evaluation boundary and rare-class recall reflects unseen defects. This result could demonstrate success for training class imbalance fold leakage but not for the stated fine-tuning image generator data rights issue. The direct test is: The training manifest contains traceable permitted sources and safety tests detect known prohibited outputs. This matters here because while preparing a production training workflow on Gemini Enterprise Agent Platform, the following condition occurs:….
D:More replicas shorten epoch time without materially changing convergence or causing network communication to dominate. This result could demonstrate success for distributed data parallel gradient synchronization but not for the stated fine-tuning image generator data rights issue. The direct test is: The training manifest contains traceable permitted sources and safety tests detect known prohibited outputs.