NCA - AI Infrastructure and Operations NCA-AIIO Practice Test 37
Question 360
A technology review accepts this control: Gate retraining output on dataset validation, safety metrics and approved promotion criteria. Which observed result provides the STRONGEST evidence of effectiveness?
- A release report includes held-out test metrics and clearly separated tuning and final evaluation datasets.
- A staging endpoint passes schema, model-load and performance smoke tests before traffic promotion.
- A sampled training dataset traces back to approved origin, version, transformation and access authorization.
- A deliberately degraded candidate is rejected while the previously approved model keeps serving.
Explanation
Correct Answer
D:The strongest proof is: A deliberately degraded candidate is rejected while the previously approved model keeps serving. It directly tests the safeguard to gate retraining output on dataset validation, safety metrics and approved promotion criteria. The control addresses this specific context: An organization automatically retrains a model nightly and publishes weights without independent acceptance rules. It matters because automated retraining is a governed change-management process, not an unconditional parameter refresh. Without it, a faulty input batch can cause an inferior model to replace a stable production version.
Incorrect Answers
A:This observation would validate a different control: Keep an appropriately isolated final test set and document evaluation metrics before selecting a release. It fits the case where a team reports model accuracy using the same data that were repeatedly used for hyperparameter selection. It cannot prove success here; instead the required evidence is a deliberately degraded candidate is rejected while the previously approved model keeps serving.
B:This observation would validate a different control: Use staged deployment validation for model artifact, endpoint compatibility, GPU resources and readiness. It fits the case where a model-serving change reaches production without checking its API schema, resource limits or startup health. It cannot prove success here; instead the required evidence is a deliberately degraded candidate is rejected while the previously approved model keeps serving.
C:This observation would validate a different control: Establish versioned data ingestion, lineage and access controls before model-development experiments. It fits the case where a model team trains on datasets whose source and consent status are not recorded across experiments. It cannot prove success here; instead the required evidence is a deliberately degraded candidate is rejected while the previously approved model keeps serving.
Question 361
An AI infrastructure team faces this requirement: Data-center management sees correct fan speeds but receives no alert when an accelerator power module reports a fault. Which action BEST addresses the operational constraint?
- Associate GPU and chassis thermal events with rack location and facility environmental telemetry.
- Scope remediation to verified affected nodes and require approval for destructive maintenance actions.
- Use governed change records, dependency checks and validated rollback procedures for cluster-wide updates.
- Integrate supported power-supply and chassis fault signals with prioritized alert routing.
Explanation
Correct Answer
D:The appropriate response is to integrate supported power-supply and chassis fault signals with prioritized alert routing. This is specific to the case where data-center management sees correct fan speeds but receives no alert when an accelerator power module reports a fault. The mechanism is that power-component health monitoring is needed even when workload throughput has not yet degraded. If ignored, one failed redundant component can remain unnoticed until a second fault causes a full outage. Validate the choice using this targeted evidence: A controlled sensor alarm reports chassis identity, severity and accountable response owner.
Incorrect Answers
A:This alternative addresses a different requirement: A training room experiences repeated thermal warnings while hardware inventory lacks rack and location mapping. It would be appropriate there to associate GPU and chassis thermal events with rack location and facility environmental telemetry. But the present scenario is about data-center management sees correct fan speeds but receives no alert when an accelerator power module reports a fault. For it, the needed control is to integrate supported power-supply and chassis fault signals with prioritized alert routing.
B:This alternative addresses a different requirement: Engineers trigger automated node remediation from broad alerts without checking device identities or active jobs. It would be appropriate there to scope remediation to verified affected nodes and require approval for destructive maintenance actions. But the present scenario is about data-center management sees correct fan speeds but receives no alert when an accelerator power module reports a fault. For it, the needed control is to integrate supported power-supply and chassis fault signals with prioritized alert routing.
C:This alternative addresses a different requirement: A team changes cluster networking and driver versions on the same day without an approval window or recovery plan. It would be appropriate there to use governed change records, dependency checks and validated rollback procedures for cluster-wide updates. But the present scenario is about data-center management sees correct fan speeds but receives no alert when an accelerator power module reports a fault. For it, the needed control is to integrate supported power-supply and chassis fault signals with prioritized alert routing.
Question 362
The selected AI infrastructure safeguard is not implemented: Track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry. The observed situation is: Research experiments change learning rates and seeds without recording configuration or resulting checkpoints. What exposure remains?
- Silent drift can degrade decisions while system CPU and GPU health still look normal.
- An accurate checkpoint can become unavailable or break client requests after release.
- Retraining may become irreproducible or rely on unauthorized data that cannot be audited.
- An apparent accuracy improvement cannot be attributed or recreated for production approval.
Explanation
Correct Answer
D:The primary exposure is that an apparent accuracy improvement cannot be attributed or recreated for production approval. This follows from research experiments change learning rates and seeds without recording configuration or resulting checkpoints. where the omitted measure was to track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry. The technical reason is that reproducible AI experiments require the parameters and input versions that produced a checkpoint. A good operational validation after repair would be: A second approved run reproduces the expected metrics within declared nondeterministic tolerances.
Incorrect Answers
A:This exposure arises in a different situation: A demand-forecasting model loses accuracy months after deployment as user behavior changes. There, the corresponding safeguard is to monitor production data distribution, predictive performance and drift with assigned retraining triggers. In the present case, the more direct risk is that an apparent accuracy improvement cannot be attributed or recreated for production approval.
B:This exposure arises in a different situation: A model-serving change reaches production without checking its API schema, resource limits or startup health. There, the corresponding safeguard is to use staged deployment validation for model artifact, endpoint compatibility, GPU resources and readiness. In the present case, the more direct risk is that an apparent accuracy improvement cannot be attributed or recreated for production approval.
C:This exposure arises in a different situation: A model team trains on datasets whose source and consent status are not recorded across experiments. There, the corresponding safeguard is to establish versioned data ingestion, lineage and access controls before model-development experiments. In the present case, the more direct risk is that an apparent accuracy improvement cannot be attributed or recreated for production approval.
Question 363
A data-center engineer must resolve the following situation: A GPU node has increasing corrected memory errors but operations waits for application crashes before investigating. What is the MOST appropriate response?
- Collect supported ECC error counters and evaluate hardware-health escalation thresholds.
- Correlate per-GPU compute, memory, scheduling contention and request tail latency.
- Correlate temperature, clock state and thermal-throttling reasons during sustained workload measurements.
- Monitor used and available GPU memory across realistic serving contexts and concurrent workload sizes.
Explanation
Correct Answer
A:The appropriate response is to collect supported ECC error counters and evaluate hardware-health escalation thresholds. This is specific to the case where a GPU node has increasing corrected memory errors but operations waits for application crashes before investigating. The mechanism is that corrected memory errors can provide early hardware-risk evidence even before visible job failure. If ignored, an emerging device fault may remain undetected until a later uncorrectable error disrupts training. Validate the choice using this targeted evidence: An injected or simulated ECC warning reaches hardware maintenance with the affected GPU identifier.
Incorrect Answers
B:This alternative addresses a different requirement: An inference service meets average latency but occasional requests become slow when multiple GPU tenants share the node. It would be appropriate there to correlate per-GPU compute, memory, scheduling contention and request tail latency. But the present scenario is about a GPU node has increasing corrected memory errors but operations waits for application crashes before investigating. For it, the needed control is to collect supported ECC error counters and evaluate hardware-health escalation thresholds.
C:This alternative addresses a different requirement: An accelerator runs at reduced clock speed during a full training epoch but device utilization appears high. It would be appropriate there to correlate temperature, clock state and thermal-throttling reasons during sustained workload measurements. But the present scenario is about a GPU node has increasing corrected memory errors but operations waits for application crashes before investigating. For it, the needed control is to collect supported ECC error counters and evaluate hardware-health escalation thresholds.
D:This alternative addresses a different requirement: A model endpoint experiences out-of-memory failures only during long prompts and concurrent sessions. It would be appropriate there to monitor used and available GPU memory across realistic serving contexts and concurrent workload sizes. But the present scenario is about a GPU node has increasing corrected memory errors but operations waits for application crashes before investigating. For it, the needed control is to collect supported ECC error counters and evaluate hardware-health escalation thresholds.
Question 364
A data-center engineer must resolve the following situation: Research experiments change learning rates and seeds without recording configuration or resulting checkpoints. What is the MOST appropriate response?
- Track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry.
- Use staged deployment validation for model artifact, endpoint compatibility, GPU resources and readiness.
- Version preprocessing and feature transformations with the model and verify training-serving parity.
- Gate retraining output on dataset validation, safety metrics and approved promotion criteria.
Explanation
Correct Answer
A:The appropriate response is to track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry. This is specific to the case where research experiments change learning rates and seeds without recording configuration or resulting checkpoints. The mechanism is that reproducible AI experiments require the parameters and input versions that produced a checkpoint. If ignored, an apparent accuracy improvement cannot be attributed or recreated for production approval. Validate the choice using this targeted evidence: A second approved run reproduces the expected metrics within declared nondeterministic tolerances.
Incorrect Answers
B:This alternative addresses a different requirement: A model-serving change reaches production without checking its API schema, resource limits or startup health. It would be appropriate there to use staged deployment validation for model artifact, endpoint compatibility, GPU resources and readiness. But the present scenario is about research experiments change learning rates and seeds without recording configuration or resulting checkpoints. For it, the needed control is to track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry.
C:This alternative addresses a different requirement: A feature-extraction job produces different input scales in training and the live inference service. It would be appropriate there to version preprocessing and feature transformations with the model and verify training-serving parity. But the present scenario is about research experiments change learning rates and seeds without recording configuration or resulting checkpoints. For it, the needed control is to track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry.
D:This alternative addresses a different requirement: An organization automatically retrains a model nightly and publishes weights without independent acceptance rules. It would be appropriate there to gate retraining output on dataset validation, safety metrics and approved promotion criteria. But the present scenario is about research experiments change learning rates and seeds without recording configuration or resulting checkpoints. For it, the needed control is to track hyperparameters, dataset versions, random seeds and artifacts in an experiment registry.
Question 365
The design team faces this case: An analytics department runs large tabular transformations on CPUs while its GPUs remain idle between training jobs. It adopts: Assess RAPIDS GPU-accelerated data processing where the data format and operators are compatible. What evidence BEST validates the intended outcome?
- A clean node pulls the approved image digest and completes a supported framework smoke test without manual dependency fixes.
- A smoke-test container successfully detects the approved GPU and executes a known CUDA kernel on the target image.
- A parity test confirms equivalent output data and measures end-to-end processing gain including transfers.
- A versioned customization experiment records the dataset, model version and before-versus-after evaluation results.
Explanation
Correct Answer
C:The strongest proof is: A parity test confirms equivalent output data and measures end-to-end processing gain including transfers. It directly tests the safeguard to assess RAPIDS GPU-accelerated data processing where the data format and operators are compatible. The control addresses this specific context: An analytics department runs large tabular transformations on CPUs while its GPUs remain idle between training jobs. It matters because gPU data-processing libraries can accelerate suitable parallel transformations without redefining the analytical objective. Without it, the preparation stage can dominate wall-clock time even when model training itself is highly accelerated.
Incorrect Answers
A:This observation would validate a different control: Use vetted NGC catalog images and pin approved container versions with tested driver compatibility. It fits the case where data scientists repeatedly download unverified deep-learning container images from personal registries before sharing GPU nodes. It cannot prove success here; instead the required evidence is a parity test confirms equivalent output data and measures end-to-end processing gain including transfers.
B:This observation would validate a different control: Confirm the supported GPU driver, CUDA runtime and application compatibility matrix before releasing a validated node image. It fits the case where a CUDA application compiled for an older environment fails to initialize after the data center deploys a newer GPU and driver image. It cannot prove success here; instead the required evidence is a parity test confirms equivalent output data and measures end-to-end processing gain including transfers.
D:This observation would validate a different control: Evaluate NVIDIA NeMo components appropriate for model customization and evaluation within the supported stack. It fits the case where an LLM customization team needs a repeatable workflow for model adaptation and responsible output evaluation instead of ad hoc scripts. It cannot prove success here; instead the required evidence is a parity test confirms equivalent output data and measures end-to-end processing gain including transfers.
Question 366
An AI infrastructure team faces this requirement: A data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. Which action BEST addresses the operational constraint?
- Assess an NVIDIA HGX platform configuration with supported GPUs, interconnect and host integration.
- Evaluate NVIDIA Spectrum Ethernet switching with appropriate RoCE and congestion-control design.
- Assess NVIDIA BlueField DPU functions for supported networking, storage and infrastructure offload.
- Consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows.
Explanation
Correct Answer
D:The appropriate response is to consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows. This is specific to the case where a data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. The mechanism is that cluster-management tools coordinate configuration and lifecycle tasks across multiple nodes rather than replacing each workload scheduler. If ignored, uncontrolled node drift can make jobs fail intermittently and slow operational recovery. Validate the choice using this targeted evidence: A test node receives the approved image and appears consistently in the inventory and health dashboard.
Incorrect Answers
A:This alternative addresses a different requirement: A server OEM plans its own data-center enclosure but wants a documented foundation for GPU compute and interconnect. It would be appropriate there to assess an NVIDIA HGX platform configuration with supported GPUs, interconnect and host integration. But the present scenario is about a data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. For it, the needed control is to consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows.
B:This alternative addresses a different requirement: A cloud architect requires an Ethernet-based accelerated AI network aligned with existing datacenter operational tooling. It would be appropriate there to evaluate NVIDIA Spectrum Ethernet switching with appropriate RoCE and congestion-control design. But the present scenario is about a data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. For it, the needed control is to consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows.
C:This alternative addresses a different requirement: An operations team wants to isolate infrastructure networking and security tasks away from host application CPUs. It would be appropriate there to assess NVIDIA BlueField DPU functions for supported networking, storage and infrastructure offload. But the present scenario is about a data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. For it, the needed control is to consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows.
Question 367
The proposed approach is to collect rack inlet and component thermal measurements during representative peak workloads. In the following setting, a rack runs multiple powerful accelerators but operators monitor only average room temperature, not intake air. Which principle BEST supports the approach?
- Local thermal conditions determine equipment behavior more directly than a remote room thermostat.
- Data-center maintenance planning must account for the coupling between compute load and cooling availability.
- High-density liquid cooling requires compatible facility infrastructure beyond server procurement.
- Redundancy only mitigates a failure if the supposedly independent paths avoid the same fault domain.
Explanation
Correct Answer
A:The governing principle is that local thermal conditions determine equipment behavior more directly than a remote room thermostat. It supports the measure to collect rack inlet and component thermal measurements during representative peak workloads. because a rack runs multiple powerful accelerators but operators monitor only average room temperature, not intake air. Without the control, one chassis can throttle or shut down without a building-wide temperature alarm. The most useful acceptance evidence is: A heat-load test records compliant inlet, GPU temperature and clock behavior for each high-density node.
Incorrect Answers
B:That principle is relevant to this distinct case: Facilities schedules maintenance to an air-handling unit while operations keeps every GPU rack at full demand. The associated response would be to coordinate workload reduction and protected shutdown windows against available cooling redundancy. It does not describe the central reason for this case, which is that local thermal conditions determine equipment behavior more directly than a remote room thermostat.
C:That principle is relevant to this distinct case: A team wants to deploy liquid-cooled GPU systems in a building that has no validated coolant distribution network. The associated response would be to check facility liquid cooling availability, heat exchangers, leak detection and vendor flow requirements. It does not describe the central reason for this case, which is that local thermal conditions determine equipment behavior more directly than a remote room thermostat.
D:That principle is relevant to this distinct case: A cluster uses redundant electrical feeds but both power cables are plugged into the same unprotected PDU. The associated response would be to validate truly independent power paths and loss-of-feed operation with qualified facility engineers. It does not describe the central reason for this case, which is that local thermal conditions determine equipment behavior more directly than a remote room thermostat.
Question 368
An AI infrastructure team faces this requirement: An industrial design group needs synchronized physics and 3D digital-twin collaboration, not a generic language-model endpoint. Which action BEST addresses the operational constraint?
- Evaluate NVIDIA Spectrum Ethernet switching with appropriate RoCE and congestion-control design.
- Evaluate the NVIDIA Quantum InfiniBand fabric against workload latency and collective communication requirements.
- Consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows.
- Evaluate NVIDIA Omniverse workflows and the required simulation and data-exchange integrations.
Explanation
Correct Answer
D:The appropriate response is to evaluate NVIDIA Omniverse workflows and the required simulation and data-exchange integrations. This is specific to the case where an industrial design group needs synchronized physics and 3D digital-twin collaboration, not a generic language-model endpoint. The mechanism is that omniverse is associated with simulation and collaborative 3D workflows rather than general LLM serving alone. If ignored, selecting only an inference endpoint would not provide the required digital-twin integration capabilities. Validate the choice using this targeted evidence: A small governed scene exchange reproduces the intended simulation state across approved collaborating tools.
Incorrect Answers
A:This alternative addresses a different requirement: A cloud architect requires an Ethernet-based accelerated AI network aligned with existing datacenter operational tooling. It would be appropriate there to evaluate NVIDIA Spectrum Ethernet switching with appropriate RoCE and congestion-control design. But the present scenario is about an industrial design group needs synchronized physics and 3D digital-twin collaboration, not a generic language-model endpoint. For it, the needed control is to evaluate NVIDIA Omniverse workflows and the required simulation and data-exchange integrations.
B:This alternative addresses a different requirement: A networking team compares technologies for low-latency all-reduce communication between training nodes. It would be appropriate there to evaluate the NVIDIA Quantum InfiniBand fabric against workload latency and collective communication requirements. But the present scenario is about an industrial design group needs synchronized physics and 3D digital-twin collaboration, not a generic language-model endpoint. For it, the needed control is to evaluate NVIDIA Omniverse workflows and the required simulation and data-exchange integrations.
C:This alternative addresses a different requirement: A data-center operator manages many accelerated nodes by hand and has inconsistent provisioning state across racks. It would be appropriate there to consider NVIDIA Base Command Manager for supported cluster provisioning, management and monitoring workflows. But the present scenario is about an industrial design group needs synchronized physics and 3D digital-twin collaboration, not a generic language-model endpoint. For it, the needed control is to evaluate NVIDIA Omniverse workflows and the required simulation and data-exchange integrations.
Question 369
The design team faces this case: A planned GPU cluster spans distant data halls but its optical cabling design ignores reach and transceiver compatibility. It adopts: Select supported optics, fiber media and loss budgets for the required data-center link distances. What evidence BEST validates the intended outcome?
- A failure-domain diagram and multi-rack load test confirm acceptable path diversity and throughput.
- Measured collective performance meets the training job's communication budget across the intended node count.
- Topology mapping confirms intra-node tensors use the expected high-bandwidth GPU interconnect.
- An optical link audit verifies approved transceiver types, signal budgets and error-free operation.
Explanation
Correct Answer
D:The strongest proof is: An optical link audit verifies approved transceiver types, signal budgets and error-free operation. It directly tests the safeguard to select supported optics, fiber media and loss budgets for the required data-center link distances. The control addresses this specific context: A planned GPU cluster spans distant data halls but its optical cabling design ignores reach and transceiver compatibility. It matters because physical optical characteristics constrain achievable port speed and link reliability. Without it, unsupported optics or excessive signal loss may cause frequent link flaps at peak utilization.
Incorrect Answers
A:This observation would validate a different control: Design an appropriate resilient leaf-spine topology and validate required bisection bandwidth. It fits the case where a data center expands the number of training racks but relies on a single aggregation switch for every east-west transfer. It cannot prove success here; instead the required evidence is an optical link audit verifies approved transceiver types, signal budgets and error-free operation.
B:This observation would validate a different control: Evaluate a suitably sized InfiniBand fabric with documented latency, bandwidth and management requirements. It fits the case where a high-performance computing group trains large models requiring fast node-to-node collective operations. It cannot prove success here; instead the required evidence is an optical link audit verifies approved transceiver types, signal budgets and error-free operation.
C:This observation would validate a different control: Use supported NVLink or NVSwitch paths for local GPU exchange rather than assuming every copy crosses Ethernet. It fits the case where a scale-up system exchanges large tensors between GPUs inside one chassis and compares that traffic with rack networking. It cannot prove success here; instead the required evidence is an optical link audit verifies approved transceiver types, signal budgets and error-free operation.
Question 370
An architecture review proposes: Include data egress, cross-region transfer and retained storage costs in the deployment decision. The underlying requirement is: A cloud-based training project moves large model checkpoints between regions and ignores transfer charges. What technical rationale MOST strongly supports it?
- Low-duty-cycle workloads often favor flexible capacity, while high continuous utilization can favor owned infrastructure.
- Infrastructure cost modeling must count data movement and persistence, not only accelerator rental time.
- Physical ownership transfers hardware and operations responsibilities to the organization.
- Workload criticality and recoverability should shape capacity purchasing choices.
Explanation
Correct Answer
B:The governing principle is that infrastructure cost modeling must count data movement and persistence, not only accelerator rental time. It supports the measure to include data egress, cross-region transfer and retained storage costs in the deployment decision. because a cloud-based training project moves large model checkpoints between regions and ignores transfer charges. Without the control, a seemingly inexpensive cloud training job may exceed budget due to repeated checkpoint transfers. The most useful acceptance evidence is: The cost estimate reconciles measured storage and transfer usage with provider billing categories.
Incorrect Answers
A:That principle is relevant to this distinct case: A small research lab runs GPU training only six days per quarter and is deciding whether to purchase a large fleet. The associated response would be to compare elastic cloud accelerator rental against on-premises ownership using realistic utilization assumptions. It does not describe the central reason for this case, which is that infrastructure cost modeling must count data movement and persistence, not only accelerator rental time.
C:That principle is relevant to this distinct case: A startup wants a fully owned GPU platform but lacks specialists to operate switching, cooling and firmware updates. The associated response would be to include operational staffing and maintenance lifecycle in the on-premises versus managed-cloud decision. It does not describe the central reason for this case, which is that infrastructure cost modeling must count data movement and persistence, not only accelerator rental time.
D:That principle is relevant to this distinct case: An AI service has predictable daily demand but a training task can be interrupted and resumed later. The associated response would be to select stable serving capacity and an appropriate lower-cost interruptible training tier with checkpoint recovery. It does not describe the central reason for this case, which is that infrastructure cost modeling must count data movement and persistence, not only accelerator rental time.