cert
cert-1
cert-2

Pass Google Professional Machine Learning Engineer Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
Professional Machine Learning Engineer Exam - Verified By Experts
Professional Machine Learning Engineer Premium Bundle
$29.98

Professional Machine Learning Engineer Premium Bundle

$79.99
$109.97
  • Premium File 402 Questions & Answers. Last update: Nov 13, 2026
  • Training Course 69 Video Lectures
  • Study Guide 376 Pages
 
$109.97
$79.99
accept 47 downloads in last 7 days
block-screenshots
Purchase Google Professional Machine Learning Engineer Exam Training Products Individually
Google Professional Machine Learning Engineer Premium File
Premium File 402 Q&A
$76.99$69.99
Google Professional Machine Learning Engineer Video Training Course
Training Course 69 Lectures
$21.99 $19.99
Google Professional Machine Learning Engineer PDF Study Guide
Study Guide 376 Pages
$21.99 $19.99
Google Professional Machine Learning Engineer Practice Test 6

Question 50

During a production ML inference release on Google Cloud, the following condition occurs: An online transformer processes one request per GPU kernel launch and throughput stays low despite steady high-frequency traffic. The service must meet its agreed quality and latency requirements. Which configuration or procedure is the most appropriate correction?

  1. Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint.
  2. Test a supported post-training quantization or optimization path and release only if target quality remains acceptable.
  3. Configure the supported private DNS, network route and endpoint access path without granting public ingress.
  4. Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency.

Explanation

Correct Answer

D:

Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency. This is the targeted corrective response for dynamic batch serving for gpu saturation. Bundling compatible requests can amortize GPU launch overhead, but excessive batch waiting raises response times. Verify the diagnosis using: Inspect batch size, kernel utilization, per-request overhead, queue wait and latency percentiles. Confirm success using: Requests per second increase while tail latency stays inside the service objective.

Incorrect Answers

A:

Choose a managed batch inference or scheduled container batch workflow instead of an always-on online endpoint. This action addresses batch versus online inference schedule, not the failure in the described dynamic batch serving for gpu saturation case. The targeted correction is: Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency.

B:

Test a supported post-training quantization or optimization path and release only if target quality remains acceptable. This action addresses inference optimization using quantization, not the failure in the described dynamic batch serving for gpu saturation case. The targeted correction is: Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency.

C:

Configure the supported private DNS, network route and endpoint access path without granting public ingress. This action addresses private vpc routing failure for serving, not the failure in the described dynamic batch serving for gpu saturation case. The targeted correction is: Enable compatible dynamic batching with a bounded wait time and benchmark throughput against p95 latency.

Question 51

During a staged restoration test at a secondary serving deployment, the following condition occurs: A lightweight CPU model needs a custom HTTP handler with sporadic traffic, but the team provisions accelerator-bound infrastructure despite no accelerator operations. The team must prove an actual user-visible improvement without weakening access controls. What should the engineer inspect before selecting a correction?

  1. Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.
  2. Inspect experiment assignment IDs, routing hashes, variant weights and per-user exposure histories.
  3. Examine scoring cadence, input size, end-to-end completion window, idle endpoint cost and per-record response requirements.
  4. Inspect idle cost, model initialization time, min replicas, burst arrival, queueing and request deadlines.

Explanation

Correct Answer

A:

Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract. This inspects evidence relevant to cloud run versus managed inference endpoint. Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model. After identifying the cause, the matching correction is: Benchmark a Cloud Run custom container against a managed Agent Platform endpoint and choose for workload fit.

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 cloud run versus managed inference endpoint. The relevant evidence is: Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.

C:

Examine scoring cadence, input size, end-to-end completion window, idle endpoint cost and per-record response requirements. This examines batch versus online inference schedule rather than the specific mechanism reported here. It is not the decisive first check for cloud run versus managed inference endpoint. The relevant evidence is: Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.

D:

Inspect idle cost, model initialization time, min replicas, burst arrival, queueing and request deadlines. This examines autoscaling minimum capacity versus cold load rather than the specific mechanism reported here. It is not the decisive first check for cloud run versus managed inference endpoint. The relevant evidence is: Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.

Question 52

During a controlled replay of the data and model handoff to a second engineering team, the following condition occurs: A churn model joins historical feature rows to today's latest cancellation status, exposing a future outcome to prior prediction dates. The handoff must preserve privacy, lineage and testable behavior. Which diagnostic check should be performed FIRST to isolate the described cause?

  1. Inspect Python runtime versions, pip dependency trees, package hashes and notebook execution metadata for both runs.
  2. Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags.
  3. Review notebook execution identity, effective IAM roles, dataset access policies and audit records for protected tables.
  4. Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer.

Explanation

Correct Answer

B:

Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags. This inspects evidence relevant to point-in-time feature leakage prevention. The globally latest value may contain future information that could not have informed the original prediction. After identifying the cause, the matching correction is: Rebuild offline training data with point-in-time eligible joins and remove features available only after each prediction moment.

Incorrect Answers

A:

Inspect Python runtime versions, pip dependency trees, package hashes and notebook execution metadata for both runs. This examines notebook dependency lock reproducibility rather than the specific mechanism reported here. It is not the decisive first check for point-in-time feature leakage prevention. The relevant evidence is: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags.

C:

Review notebook execution identity, effective IAM roles, dataset access policies and audit records for protected tables. This examines colab enterprise notebook data access scope rather than the specific mechanism reported here. It is not the decisive first check for point-in-time feature leakage prevention. The relevant evidence is: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags.

D:

Compare entity IDs, key types, online feature-view definitions, offline joins and records for the same verified customer. This examines feature store online and offline entity alignment rather than the specific mechanism reported here. It is not the decisive first check for point-in-time feature leakage prevention. The relevant evidence is: Inspect feature event timestamps, prediction cutoffs, feature joins and any post-cancellation administrative flags.

Question 53

As the team tests a new governed AI release against adversarial and normal data, the following condition occurs: An approval model shows unchanged overall AUC but a recent feature change sharply increases false negatives for a protected subgroup. The fix must retain legitimate model usefulness and appropriate controls. Which statement identifies the primary reason for this failure?

  1. Aggregate model metrics can conceal uneven errors and harms across different population segments.
  2. Screening helps detect malicious instructions or sensitive content crossing the model trust boundary.
  3. Sensitive information should be shared only when necessary and permitted; ordinary prompting is not an exemption from privacy controls.
  4. Fluent text or semantic similarity is not proof an answer is supported by the retrieved source material.

Explanation

Correct Answer

A:

Aggregate model metrics can conceal uneven errors and harms across different population segments. This explains the mechanism behind model feature bias by subgroup and why the described failure is possible. A discriminating investigation is: Inspect subgroup confusion matrices, sample sizes, training representation and evaluation slices over time. The corresponding repair is: Evaluate and address documented subgroup disparities using appropriate fairness metrics and approved data/model changes.

Incorrect Answers

B:

Screening helps detect malicious instructions or sensitive content crossing the model trust boundary. This principle describes model armor untrusted content screening, which is not the cause of the observed model feature bias by subgroup failure. The controlling explanation is: Aggregate model metrics can conceal uneven errors and harms across different population segments.

C:

Sensitive information should be shared only when necessary and permitted; ordinary prompting is not an exemption from privacy controls. This principle describes sensitive identifiers leaked to llm prompts, which is not the cause of the observed model feature bias by subgroup failure. The controlling explanation is: Aggregate model metrics can conceal uneven errors and harms across different population segments.

D:

Fluent text or semantic similarity is not proof an answer is supported by the retrieved source material. This principle describes generative ai factual grounding evaluation, which is not the cause of the observed model feature bias by subgroup failure. The controlling explanation is: Aggregate model metrics can conceal uneven errors and harms across different population segments.

Question 54

During a regulated MLOps pipeline deployment, 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 workflow must be repeatable and protect the last approved production model. Which statement identifies the primary reason for this failure?

  1. Evaluation without a blocking release decision cannot protect production from quality regressions.
  2. Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries.
  3. Promoting immutable pipeline definitions prevents untracked hand edits from changing processing semantics.
  4. Retraining on a timer alone does not imply better models and may use incomplete or unrepresentative labels.

Explanation

Correct Answer

B:

Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries. This explains the mechanism behind serving schema validation component and why the described failure is possible. A discriminating investigation is: Inspect incoming schema versions, validation logs, casting rules and downstream prediction anomalies. The corresponding repair is: Enforce a shared typed schema contract and reject incompatible feature versions before training or inference.

Incorrect Answers

A:

Evaluation without a blocking release decision cannot protect production from quality regressions. This principle describes model evaluation gate before deployment, which is not the cause of the observed serving schema validation component failure. The controlling explanation is: Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries.

C:

Promoting immutable pipeline definitions prevents untracked hand edits from changing processing semantics. This principle describes pipeline environment parameter promotion, which is not the cause of the observed serving schema validation component failure. The controlling explanation is: Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries.

D:

Retraining on a timer alone does not imply better models and may use incomplete or unrepresentative labels. This principle describes data drift triggered retraining policy, which is not the cause of the observed serving schema validation component failure. The controlling explanation is: Silent casting can hide producer errors and produce invalid model inputs across pipeline boundaries.

Question 55

During a staged restoration test at a secondary serving deployment, the following condition occurs: A lightweight CPU model needs a custom HTTP handler with sporadic traffic, but the team provisions accelerator-bound infrastructure despite no accelerator operations. The team must prove an actual user-visible improvement without weakening access controls. Which statement identifies the primary reason for this failure?

  1. Mutable tags can point to new image content, defeating reproducibility even if the deployment text is unchanged.
  2. Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model.
  3. Mutable nicknames do not establish immutable model provenance for controlled production releases.
  4. CPU cores do not expand device memory; accelerator selection must match model footprint and request shape.

Explanation

Correct Answer

B:

Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model. This explains the mechanism behind cloud run versus managed inference endpoint and why the described failure is possible. A discriminating investigation is: Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract. The corresponding repair is: Benchmark a Cloud Run custom container against a managed Agent Platform endpoint and choose for workload fit.

Incorrect Answers

A:

Mutable tags can point to new image content, defeating reproducibility even if the deployment text is unchanged. This principle describes model artifact image digest reproducibility, which is not the cause of the observed cloud run versus managed inference endpoint failure. The controlling explanation is: Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model.

C:

Mutable nicknames do not establish immutable model provenance for controlled production releases. This principle describes model registry version provenance, which is not the cause of the observed cloud run versus managed inference endpoint failure. The controlling explanation is: Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model.

D:

CPU cores do not expand device memory; accelerator selection must match model footprint and request shape. This principle describes gpu cost and memory fit for large embedding model, which is not the cause of the observed cloud run versus managed inference endpoint failure. The controlling explanation is: Containerized CPU inference can be appropriate on serverless compute; specialized ML endpoints are not required for every model.

Question 56

Within a jointly maintained Google Cloud data science workspace, the following condition occurs: A model evaluation cell appears to pass only because an old variable remains in kernel memory after cells were executed out of order. Collaborators must avoid expanding data access or breaking experiment reproducibility. Which statement identifies the primary reason for this failure?

  1. Partition pruning avoids irrelevant I/O when predicates match the storage partitioning scheme.
  2. Notebook execution can retain stale objects; a clean deterministic run detects hidden state dependencies.
  3. Framework-native checkpoints and runtime APIs are not automatically interchangeable across ML libraries.
  4. Single-machine notebook memory is limited; warehouse operations should execute near large warehouse-resident data.

Explanation

Correct Answer

B:

Notebook execution can retain stale objects; a clean deterministic run detects hidden state dependencies. This explains the mechanism behind notebook transient state and execution order and why the described failure is possible. A discriminating investigation is: Restart the runtime, run the notebook top to bottom and compare variable initialization with committed source revision. The corresponding repair is: Refactor hidden state into explicit inputs and functions and add a clean-kernel end-to-end execution gate.

Incorrect Answers

A:

Partition pruning avoids irrelevant I/O when predicates match the storage partitioning scheme. This principle describes bigquery partition filter cost optimization, which is not the cause of the observed notebook transient state and execution order failure. The controlling explanation is: Notebook execution can retain stale objects; a clean deterministic run detects hidden state dependencies.

C:

Framework-native checkpoints and runtime APIs are not automatically interchangeable across ML libraries. This principle describes model garden jax accelerator compatibility, which is not the cause of the observed notebook transient state and execution order failure. The controlling explanation is: Notebook execution can retain stale objects; a clean deterministic run detects hidden state dependencies.

D:

Single-machine notebook memory is limited; warehouse operations should execute near large warehouse-resident data. This principle describes notebook memory exhaustion on full data extraction, which is not the cause of the observed notebook transient state and execution order failure. The controlling explanation is: Notebook execution can retain stale objects; a clean deterministic run detects hidden state dependencies.

Question 57

During a staged restoration test at a secondary serving deployment, the following condition occurs: An internal fraud scorer sends sensitive feature vectors to a publicly accessible prediction endpoint even though clients are inside a controlled VPC. 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?

  1. Inspect experiment assignment IDs, routing hashes, variant weights and per-user exposure histories.
  2. Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.
  3. Inspect memory footprint, concurrency, request arrival, cold-start tolerance and supported serving framework contract.
  4. Inspect feature-view sync status, latest source timestamps, entity ID and online lookup freshness.

Explanation

Correct Answer

B:

Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements. This inspects evidence relevant to private endpoint for regulated inference traffic. Network exposure and IAM controls are complementary; a public endpoint adds attack surface even with authentication. After identifying the cause, the matching correction is: Deploy a supported private endpoint or constrained private network path with least-privilege caller authorization.

Incorrect Answers

A:

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 private endpoint for regulated inference traffic. The relevant evidence is: Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.

C:

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 private endpoint for regulated inference traffic. The relevant evidence is: Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.

D:

Inspect feature-view sync status, latest source timestamps, entity ID and online lookup freshness. This examines feature store freshness and online sync rather than the specific mechanism reported here. It is not the decisive first check for private endpoint for regulated inference traffic. The relevant evidence is: Examine endpoint exposure settings, caller IAM, VPC routes, private connectivity and data egress requirements.

Question 58

At a Google Cloud analytics team preparing a governed pilot, 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 team must satisfy a reproducible low-code AI assessment. What acceptance result verifies the intended behavior after remediation?

  1. Classification F1 remains above the acceptance threshold while end-to-end latency and cost per thousand requests fall.
  2. Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values.
  3. A controlled raw record yields the same transformed feature representation for evaluation and production SQL inference.
  4. Stable clusters reveal coherent customer behavior without pretending an unsupervised assignment is known supervised ground truth.

Explanation

Correct Answer

B:

Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values. This is specific acceptance evidence for bigquery ml linear regression for continuous value. Classification labels cannot directly represent unrestricted continuous outcomes; SQL regression models match numeric targets. It confirms the controlled correction: Use a BigQuery ML LINEAR_REG or suitable regression model with a numeric target and evaluate continuous-error metrics on held-out records.

Incorrect Answers

A:

Classification F1 remains above the acceptance threshold while end-to-end latency and cost per thousand requests fall. This result could demonstrate success for gemini model choice for latency versus complexity but not for the stated bigquery ml linear regression for continuous value issue. The direct test is: Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values.

C:

A controlled raw record yields the same transformed feature representation for evaluation and production SQL inference. This result could demonstrate success for bigquery ml transform serving parity but not for the stated bigquery ml linear regression for continuous value issue. The direct test is: Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values.

D:

Stable clusters reveal coherent customer behavior without pretending an unsupervised assignment is known supervised ground truth. This result could demonstrate success for bigquery ml kmeans customer segmentation but not for the stated bigquery ml linear regression for continuous value issue. The direct test is: Predictions contain sensible numeric estimates and holdout MAE or RMSE is measured against actual monetary values.

Question 59

As the team tests a new governed AI release against adversarial and normal data, the following condition occurs: A model keeps similar average scores after an upstream data migration, but feature importance shifts from stable signals toward a suspicious proxy variable. The fix must retain legitimate model usefulness and appropriate controls. Which TWO complementary choices jointly establish the evidence and appropriate correction? Select TWO.

  1. Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.
  2. Apply supported safety filters and output policy checks with a controlled refusal or human review path for blocked results.
  3. Inspect attachment ingestion, trust boundaries, malicious prompt examples and deployed Model Armor filter results.
  4. Require validated business quality and technical reliability thresholds to continue rollout, with tested rollback on failure.
  5. Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort.

Explanation

Correct Answers

A:

Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation. This is the targeted corrective response for feature attribution drift after deployment. Feature attribution drift can indicate changing model reliance even when overall score distributions remain stable. Verify the diagnosis using: Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. Confirm success using: Attribution changes are explained or remediated and decision drivers remain stable on known reference cases.

E:

Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. This inspects evidence relevant to feature attribution drift after deployment. Feature attribution drift can indicate changing model reliance even when overall score distributions remain stable. After identifying the cause, the matching correction is: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

Incorrect Answers

B:

Apply supported safety filters and output policy checks with a controlled refusal or human review path for blocked results. This is an action or diagnostic relevant to gen ai output safety policy enforcement, not one of the required two steps. In this scenario verify: Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. Then correct: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

C:

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: Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. Then correct: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

D:

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: Examine explanation distributions, feature generation changes, protected attribute proxies and prediction outcomes by cohort. Then correct: Investigate attribution drift, remove invalid proxy features if warranted and retrain with appropriate validation.

Question 60

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 diagnostic check should be performed FIRST to isolate the described cause?

  1. Inspect attachment ingestion, trust boundaries, malicious prompt examples and deployed Model Armor filter results.
  2. Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.
  3. Review prompt token counts, output length, cache behavior, p95 model latency and cost per successful task.
  4. Review the business safety taxonomy, representative adversarial evaluation prompts, safety-filter configuration and escalation design.

Explanation

Correct Answer

B:

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.

Incorrect Answers

A:

Inspect attachment ingestion, trust boundaries, malicious prompt examples and deployed Model Armor filter results. This examines model armor untrusted content screening rather than the specific mechanism reported here. It is not the decisive first check for prompt injection through retrieved documents. The relevant evidence is: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.

C:

Review prompt token counts, output length, cache behavior, p95 model latency and cost per successful task. This examines llm latency and cost token regression rather than the specific mechanism reported here. It is not the decisive first check for prompt injection through retrieved documents. The relevant evidence is: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.

D:

Review the business safety taxonomy, representative adversarial evaluation prompts, safety-filter configuration and escalation design. This examines gen ai output safety policy enforcement rather than the specific mechanism reported here. It is not the decisive first check for prompt injection through retrieved documents. The relevant evidence is: Inspect retrieved document boundaries, instruction hierarchy, malicious test prompts and the application's output actions.

Professional Machine Learning Engineer Practice Tests