Question 650A startup requires a serverless prediction API for a lightweight CPU-based model with modest cold-start tolerance. What should the team change to address the specific technical requirement? Package and version inference preprocessing and postprocessing together with the approved model artifact. Choose serving accelerator memory and concurrency limits based on measured token or batch demand. Run a randomized A/B test with stable assignment, clear primary metrics and sufficient sample duration. Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed. ExplanationCorrect Answer D:Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. To validate the correction: Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. Other Answers A:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. It does not resolve the specific failure; the targeted correction is: Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed. B:Generative inference memory depends on model parameters, KV-cache and concurrent sequence lengths, not only average CPU use. It does not resolve the specific failure; the targeted correction is: Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed. C:Randomized model assignment estimates causal effects more reliably than comparing different uncontrolled traffic windows. It does not resolve the specific failure; the targeted correction is: Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed.
Question 651A startup requires a serverless prediction API for a lightweight CPU-based model with modest cold-start tolerance. Which underlying ML principle makes the proposed approach effective? Hardware acceleration is beneficial only when supported operators and workload intensity justify the resource premium. Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. ExplanationCorrect Answer B:Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. This supports the specific implementation: Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed. A suitable independent check is: Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. Other Answers A:Hardware acceleration is beneficial only when supported operators and workload intensity justify the resource premium. The cause in the stated scenario is different: Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. C:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. The cause in the stated scenario is different: Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. D:Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. The cause in the stated scenario is different: Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter.
Question 652A startup requires a serverless prediction API for a lightweight CPU-based model with modest cold-start tolerance. What is the most relevant acceptance check for the proposed ML change? Verify complete output row counts, cost per prediction and batch completion within the required business window. Measure cold-start frequency, p95 first-token delay and monthly idle cost with and without warm capacity. Load test p95 latency, error rate and scaling behavior under expected concurrent traffic. Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. ExplanationCorrect Answer D:Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. This tests the underlying failure: Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. Other Answers A:Batch prediction amortizes job start overhead across large datasets and avoids always-on serving resources. That does not establish the required outcome; verify it by checking: Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. B:Warm replicas reduce first-request initialization latency but incur charges while receiving no traffic. That does not establish the required outcome; verify it by checking: Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior. C:Online prediction keeps inference capacity available for request-response workloads at the cost of capacity management. That does not establish the required outcome; verify it by checking: Test first-request latency, sustained throughput, request authentication and scale-to-zero behavior.
Question 653A startup requires a serverless prediction API for a lightweight CPU-based model with modest cold-start tolerance. What preventive release checklist item best protects the intended outcome? Pre-register success metrics, sample size and stop criteria to limit experiment selection bias. Monitor cold starts and cost per request and document when dedicated model endpoints become necessary. Require a documented latency justification before reserving always-on warm serving resources. Make preprocessing parity and schema-contract checks mandatory in model rollout pipelines. ExplanationCorrect Answer B:Monitor cold starts and cost per request and document when dedicated model endpoints become necessary. This protects the selected approach because Cloud Run scales container services with requests and may reduce idle cost, though cold starts and resource limits matter. The immediate solution is: Evaluate Cloud Run container serving and configure concurrency, authentication and minimum instances as needed. Other Answers A:Randomized model assignment estimates causal effects more reliably than comparing different uncontrolled traffic windows. The continuing safeguard required here is: Monitor cold starts and cost per request and document when dedicated model endpoints become necessary. C:Warm replicas reduce first-request initialization latency but incur charges while receiving no traffic. The continuing safeguard required here is: Monitor cold starts and cost per request and document when dedicated model endpoints become necessary. D:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. The continuing safeguard required here is: Monitor cold starts and cost per request and document when dedicated model endpoints become necessary.
Question 654A team must serve an existing specialized predictor with complex sidecars and custom network policy. Which proposed approach best resolves the described model engineering problem? Profile per-request token generation, model size and decoding settings before scaling only the replica count. Register versioned model artifacts in Gemini Enterprise Agent Platform Model Registry and deploy approved versions by ID. Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. Build a minimal supported custom serving container with the required runtime, prediction handler and health probes. ExplanationCorrect Answer C:Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. To validate the correction: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. Other Answers A:Horizontal scaling raises concurrent capacity but does not necessarily accelerate computation within an individual inference request. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. B:Model registry versioning connects a deployable model to metadata, approval and rollback history. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. D:A custom container controls dependencies and request handling but inherits security, compatibility and patching duties. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management.
Question 655A team must serve an existing specialized predictor with complex sidecars and custom network policy. Which property of the selected ML workflow accounts for this requirement? Batch prediction amortizes job start overhead across large datasets and avoids always-on serving resources. Scaling additional inference replicas can improve throughput but cold-start and quota limitations constrain capacity changes. Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. ExplanationCorrect Answer D:Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. This supports the specific implementation: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. A suitable independent check is: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. Other Answers A:Batch prediction amortizes job start overhead across large datasets and avoids always-on serving resources. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. B:Scaling additional inference replicas can improve throughput but cold-start and quota limitations constrain capacity changes. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. C:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations.
Question 656A team must serve an existing specialized predictor with complex sidecars and custom network policy. What controlled validation most directly confirms the approach worked? Compare per-version request counts, quality, error rate and latency before raising candidate traffic share. Run smoke and load tests against the container image digest and confirm readiness and error handling. Test load ramps and observe latency, replica counts and saturation before and after scaling configuration. Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. ExplanationCorrect Answer D:Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. This tests the underlying failure: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. Other Answers A:Canary traffic allocation limits the initial exposure of a new model while collecting live performance evidence. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. B:A custom container controls dependencies and request handling but inherits security, compatibility and patching duties. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. C:Scaling additional inference replicas can improve throughput but cold-start and quota limitations constrain capacity changes. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance.
Question 657A team must serve an existing specialized predictor with complex sidecars and custom network policy. What governance or operational safeguard should accompany the implementation? Keep container image policies, pod security and service-mesh observability under centralized platform ownership. Make preprocessing parity and schema-contract checks mandatory in model rollout pipelines. Pre-register success metrics, sample size and stop criteria to limit experiment selection bias. Review endpoint exposure and identity policies before enabling any new inference consumer. ExplanationCorrect Answer A:Keep container image policies, pod security and service-mesh observability under centralized platform ownership. This protects the selected approach because Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. The immediate solution is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. Other Answers B:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. The continuing safeguard required here is: Keep container image policies, pod security and service-mesh observability under centralized platform ownership. C:Randomized model assignment estimates causal effects more reliably than comparing different uncontrolled traffic windows. The continuing safeguard required here is: Keep container image policies, pod security and service-mesh observability under centralized platform ownership. D:Endpoint network reachability and cloud identity authorization are separate security boundaries. The continuing safeguard required here is: Keep container image policies, pod security and service-mesh observability under centralized platform ownership.
Question 658A regulated service already has GKE operational capacity and requires Kubernetes-native inference controls. Which Google Cloud machine-learning design choice best satisfies the stated requirement? Run a randomized A/B test with stable assignment, clear primary metrics and sufficient sample duration. Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. Register versioned model artifacts in Gemini Enterprise Agent Platform Model Registry and deploy approved versions by ID. Deploy a supported private endpoint or private networking controls with least-privilege service authentication. ExplanationCorrect Answer B:Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. To validate the correction: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. Other Answers A:Randomized model assignment estimates causal effects more reliably than comparing different uncontrolled traffic windows. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. C:Model registry versioning connects a deployable model to metadata, approval and rollback history. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. D:Endpoint network reachability and cloud identity authorization are separate security boundaries. It does not resolve the specific failure; the targeted correction is: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management.
Question 659A regulated service already has GKE operational capacity and requires Kubernetes-native inference controls. Which underlying ML principle makes the proposed approach effective? A custom container controls dependencies and request handling but inherits security, compatibility and patching duties. Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. Scaling additional inference replicas can improve throughput but cold-start and quota limitations constrain capacity changes. ExplanationCorrect Answer C:Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. This supports the specific implementation: Deploy a governed containerized model service on GKE when custom runtime and policy requirements justify cluster management. A suitable independent check is: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. Other Answers A:A custom container controls dependencies and request handling but inherits security, compatibility and patching duties. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. B:Training-serving feature skew can arise when model inputs are transformed inconsistently across environments. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. D:Scaling additional inference replicas can improve throughput but cold-start and quota limitations constrain capacity changes. The cause in the stated scenario is different: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations.
Question 660A regulated service already has GKE operational capacity and requires Kubernetes-native inference controls. Which measurement or test should the team perform after implementation? Deploy a test artifact and confirm request schema, predictions and runtime library compatibility. Measure cold-start frequency, p95 first-token delay and monthly idle cost with and without warm capacity. Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. Run smoke and load tests against the container image digest and confirm readiness and error handling. ExplanationCorrect Answer C:Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. This tests the underlying failure: Kubernetes-based serving supports flexible sidecars and orchestration while increasing operational maintenance obligations. Other Answers A:Supported prebuilt serving environments reduce runtime management but depend on matching artifact formats and versions. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. B:Warm replicas reduce first-request initialization latency but incur charges while receiving no traffic. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance. D:A custom container controls dependencies and request handling but inherits security, compatibility and patching duties. That does not establish the required outcome; verify it by checking: Inspect horizontal scaling, readiness checks, network-policy enforcement and per-pod performance.