Microsoft AI-102 (Designing and Implementing a Microsoft Azure AI Solution) Practice Test 7

Question 60

A model cannot be deployed even though a similar family was deployed last quarter in a different region. What should the team confirm?

  1. Verify regional support for the exact model version and deployment category
  2. Grant the configured identity or credential the required permissions on the storage resource
  3. Review the model's licensing terms and allowed deployment offer
  4. Review project-scoped access assignments for that identity

Explanation

Correct Answer

A:

Availability can depend on model, version, and offer type; previous success in another region is not a universal guarantee. In the enterprise generative-AI development environment, confirm the fix by retesting the affected behavior after verifying regional support for the exact model version and deployment category.

Incorrect Answers

B:

A connection records how to reach a service but does not replace its authorization checks in its own workflow, but that is a different correction from the present requirement to verify regional support for the exact model version and deployment category. The reported issue would remain.

C:

That change addresses another failure mode: model catalogs include different publishers and offers; capability alone does not establish permitted commercial use; it does not address this case, for which the relevant intervention is to verify regional support for the exact model version and deployment category.

D:

Although useful elsewhere in the enterprise generative-AI development environment, environment access does not necessarily authorize every project action; permissions must match the target project scope; the present case is different and calls for the team to verify regional support for the exact model version and deployment category.

Question 61

A team wants factual answers about frequently changing HR policies and proposes supervised fine-tuning. What should it do instead?

  1. Retrieve current approved policy content at query time through grounding
  2. Target the deployment configured with the purchased provisioned throughput
  3. Verify the container's supported connectivity and metering requirements before designing for offline operation
  4. Track the batch job identifier and retrieve results after asynchronous completion

Explanation

Correct Answer

A:

Fine-tuning adjusts model behavior, but is not a dependable replacement for up-to-date factual retrieval from mutable documents. In the production inference rollout, confirm the fix by retesting the affected behavior after retrieving current approved policy content at query time through grounding.

Incorrect Answers

B:

Although useful elsewhere in the production inference rollout, the capacity reservation applies to the supported provisioned deployment, not automatically to every deployment in the resource; the present case is different and calls for the team to retrieve current approved policy content at query time through grounding.

C:

That change addresses another failure mode: supported containers have licensing, network, and usage-accounting conditions; not every service supports disconnected operation; it does not address this case, for which the relevant intervention is to retrieve current approved policy content at query time through grounding.

D:

Batch processing is not a synchronous completion API; the client should poll job state or use supported completion notification in its own workflow, but that is a different correction from the present requirement to retrieve current approved policy content at query time through grounding. The reported issue would remain.

Question 62

The customer demands a response within 150 milliseconds for a chat turn. Which processing choice is inappropriate?

  1. Compare sustained token demand and latency targets with the provisioned capacity allocation
  2. Treat standard request capacity as subject to quota and service limits, not a reserved guarantee
  3. Profile container CPU, memory, and supported accelerator needs using realistic documents
  4. Do not route the latency-critical request through an offline batch job

Explanation

Correct Answer

D:

Batch queues target throughput and cost objectives rather than interactive request-response latency. In the production inference rollout, confirm the fix by retesting the affected behavior after doing not route the latency-critical request through an offline batch job.

Incorrect Answers

A:

That change addresses another failure mode: provisioned units bound allocated throughput; capacity planning must account for observed request and token mix; it does not address this case, for which the relevant intervention is to do not route the latency-critical request through an offline batch job.

B:

Standard online serving can be throttled under limits; guaranteed dedicated capacity requires an appropriate supported offering in its own workflow, but that is a different correction from the present requirement to do not route the latency-critical request through an offline batch job. The reported issue would remain.

C:

Although useful elsewhere in the production inference rollout, container inference throughput depends on the host resources and image requirements as well as network conditions; the present case is different and calls for the team to do not route the latency-critical request through an offline batch job.

Question 63

The assistant has intermittent low-volume traffic and no contractual need for reserved model capacity. Which cost model is more appropriate?

  1. Retrieve current approved policy content at query time through grounding
  2. Target the deployment configured with the purchased provisioned throughput
  3. Verify the container's supported connectivity and metering requirements before designing for offline operation
  4. Use a compatible standard online deployment with usage-based billing

Explanation

Correct Answer

D:

Standard deployments typically support variable interactive demand without reserving provisioned serving units. In the production inference rollout, confirm the fix by retesting the affected behavior after using a compatible standard online deployment with usage-based billing.

Incorrect Answers

A:

Fine-tuning adjusts model behavior, but is not a dependable replacement for up-to-date factual retrieval from mutable documents in its own workflow, but that is a different correction from the present requirement to use a compatible standard online deployment with usage-based billing. The reported issue would remain.

B:

Although useful elsewhere in the production inference rollout, the capacity reservation applies to the supported provisioned deployment, not automatically to every deployment in the resource; the present case is different and calls for the team to use a compatible standard online deployment with usage-based billing.

C:

That change addresses another failure mode: supported containers have licensing, network, and usage-accounting conditions; not every service supports disconnected operation; it does not address this case, for which the relevant intervention is to use a compatible standard online deployment with usage-based billing.

Question 64

A container is deployed successfully but cannot load the requested hosted proprietary chat model. What is the architectural issue?

  1. Deploy and target the fine-tuned model artifact for inference
  2. Validate model, region, and deployment-type eligibility for provisioned throughput
  3. Confirm the particular AI feature is actually distributed as a supported container image
  4. Inspect per-record errors and retry only eligible failed records after correction

Explanation

Correct Answer

C:

Containers are available only for selected services and models; a running container host does not imply every cloud inference model can run locally. In the production inference rollout, confirm the fix by retesting the affected behavior after confirming the particular AI feature is actually distributed as a supported container image.

Incorrect Answers

A:

Although useful elsewhere in the production inference rollout, training does not automatically redirect existing inference traffic; clients must address the fine-tuned deployment; the present case is different and calls for the team to confirm the particular AI feature is actually distributed as a supported container image.

B:

Provisioned availability varies; reserving units should follow compatibility and sizing checks, not model-name assumptions in its own workflow, but that is a different correction from the present requirement to confirm the particular AI feature is actually distributed as a supported container image. The reported issue would remain.

D:

That change addresses another failure mode: asynchronous bulk jobs can yield partial failures; successful records need not be blindly reprocessed; it does not address this case, for which the relevant intervention is to confirm the particular AI feature is actually distributed as a supported container image.

Question 65

A large batch request was submitted successfully, but the calling app times out while waiting synchronously. What should the client do?

  1. Track the batch job identifier and retrieve results after asynchronous completion
  2. Retrieve current approved policy content at query time through grounding
  3. Use a compatible standard online deployment with usage-based billing
  4. Verify the container's supported connectivity and metering requirements before designing for offline operation

Explanation

Correct Answer

A:

Batch processing is not a synchronous completion API; the client should poll job state or use supported completion notification. In the production inference rollout, confirm the fix by retesting the affected behavior after tracking the batch job identifier and retrieve results after asynchronous completion.

Incorrect Answers

B:

Although useful elsewhere in the production inference rollout, fine-tuning adjusts model behavior, but is not a dependable replacement for up-to-date factual retrieval from mutable documents; the present case is different and calls for the team to track the batch job identifier and retrieve results after asynchronous completion.

C:

That change addresses another failure mode: standard deployments typically support variable interactive demand without reserving provisioned serving units; it does not address this case, for which the relevant intervention is to track the batch job identifier and retrieve results after asynchronous completion.

D:

Supported containers have licensing, network, and usage-accounting conditions; not every service supports disconnected operation in its own workflow, but that is a different correction from the present requirement to track the batch job identifier and retrieve results after asynchronous completion. The reported issue would remain.

Question 66

A team wants to purchase capacity for a new model version but does not know whether it supports provisioning. What must be checked?

  1. Move eligible large offline jobs to a supported batch inference path
  2. Inspect per-record errors and retry only eligible failed records after correction
  3. Validate model, region, and deployment-type eligibility for provisioned throughput
  4. Deploy and target the fine-tuned model artifact for inference

Explanation

Correct Answer

C:

Provisioned availability varies; reserving units should follow compatibility and sizing checks, not model-name assumptions. In the production inference rollout, confirm the fix by retesting the affected behavior after validating model, region, and deployment-type eligibility for provisioned throughput.

Incorrect Answers

A:

Asynchronous batch work is better suited to queued bulk inputs; sharing the interactive path can create avoidable contention in its own workflow, but that is a different correction from the present requirement to validate model, region, and deployment-type eligibility for provisioned throughput. The reported issue would remain.

B:

That change addresses another failure mode: asynchronous bulk jobs can yield partial failures; successful records need not be blindly reprocessed; it does not address this case, for which the relevant intervention is to validate model, region, and deployment-type eligibility for provisioned throughput.

D:

Although useful elsewhere in the production inference rollout, training does not automatically redirect existing inference traffic; clients must address the fine-tuned deployment; the present case is different and calls for the team to validate model, region, and deployment-type eligibility for provisioned throughput.

Question 67

Traffic repeatedly exceeds the purchased capacity during peak hours. What planning input needs review?

  1. Treat standard request capacity as subject to quota and service limits, not a reserved guarantee
  2. Compare sustained token demand and latency targets with the provisioned capacity allocation
  3. Validate dataset rights, minimization, and the supported fine-tuning data-handling policy
  4. Profile container CPU, memory, and supported accelerator needs using realistic documents

Explanation

Correct Answer

B:

Provisioned units bound allocated throughput; capacity planning must account for observed request and token mix. In the production inference rollout, confirm the fix by retesting the affected behavior after comparing sustained token demand and latency targets with the provisioned capacity allocation.

Incorrect Answers

A:

Although useful elsewhere in the production inference rollout, standard online serving can be throttled under limits; guaranteed dedicated capacity requires an appropriate supported offering; the present case is different and calls for the team to compare sustained token demand and latency targets with the provisioned capacity allocation.

C:

A model's ability to fine-tune does not authorize use of confidential or unlicensed training material in its own workflow, but that is a different correction from the present requirement to compare sustained token demand and latency targets with the provisioned capacity allocation. The reported issue would remain.

D:

That change addresses another failure mode: container inference throughput depends on the host resources and image requirements as well as network conditions; it does not address this case, for which the relevant intervention is to compare sustained token demand and latency targets with the provisioned capacity allocation.

Question 68

A manufacturing site loses internet connectivity periodically, but wants to run an Azure AI container locally. What prerequisite needs validation?

  1. Use a compatible standard online deployment with usage-based billing
  2. Verify the container's supported connectivity and metering requirements before designing for offline operation
  3. Retrieve current approved policy content at query time through grounding
  4. Target the deployment configured with the purchased provisioned throughput

Explanation

Correct Answer

B:

Supported containers have licensing, network, and usage-accounting conditions; not every service supports disconnected operation. In the production inference rollout, confirm the fix by retesting the affected behavior after verifying the container's supported connectivity and metering requirements before designing for offline operation.

Incorrect Answers

A:

That change addresses another failure mode: standard deployments typically support variable interactive demand without reserving provisioned serving units; it does not address this case, for which the relevant intervention is to verify the container's supported connectivity and metering requirements before designing for offline operation.

C:

Although useful elsewhere in the production inference rollout, fine-tuning adjusts model behavior, but is not a dependable replacement for up-to-date factual retrieval from mutable documents; the present case is different and calls for the team to verify the container's supported connectivity and metering requirements before designing for offline operation.

D:

The capacity reservation applies to the supported provisioned deployment, not automatically to every deployment in the resource in its own workflow, but that is a different correction from the present requirement to verify the container's supported connectivity and metering requirements before designing for offline operation. The reported issue would remain.

Question 69

An offline batch inference job produced output for most input records but reports errors for some records. Which TWO actions should the recovery workflow perform before considering the dataset complete? (Choose TWO.)

  1. Retry only the eligible failed records after correcting their underlying issues
  2. Identify which input records failed and capture their error details
  3. Validate model, region, and deployment-type eligibility for provisioned throughput
  4. Deploy and target the fine-tuned model artifact for inference
  5. Confirm the particular AI feature is actually distributed as a supported container image

Explanation

Correct Answers

A:

Selective retries avoid duplicate processing and let the team verify that each previously failed record now has a valid result.

B:

Per-record job statuses distinguish successful output from individual failures, permitting targeted recovery without repeating records that already completed.

Incorrect Answers

C:

Although useful elsewhere in the production inference rollout, provisioned availability varies; reserving units should follow compatibility and sizing checks, not model-name assumptions; the present case is different and calls for the team to identify which input records failed and capture their error details and retry only the eligible failed records after correcting their underlying issues.

D:

Training does not automatically redirect existing inference traffic; clients must address the fine-tuned deployment in its own workflow, but that is a different correction from the present requirement to identify which input records failed and capture their error details and retry only the eligible failed records after correcting their underlying issues. The reported issue would remain.

E:

That change addresses another failure mode: containers are available only for selected services and models; a running container host does not imply every cloud inference model can run locally; it does not address this case, for which the relevant intervention is to identify which input records failed and capture their error details and retry only the eligible failed records after correcting their underlying issues.

Question 70

In a production inference rollout, engineers must run a supported AI capability close to data in a controlled container host. The expected evidence is that the workload runs on the chosen container host with the required metering setup. Which implementation is the best match?

  1. Use a batch-processing deployment or API
  2. Deploy a supported Azure AI container
  3. Select a standard online model deployment
  4. Select a provisioned-throughput model deployment
  5. Fine-tune a supported model before deployment

Explanation

Correct Answer

B:

Containers can bring supported Azure AI inference features to an environment that meets their licensing and connectivity conditions. For this production inference rollout, the intended evidence is that the workload runs on the chosen container host with the required metering setup. Not every model or Foundry service is available as a container; compatibility must be checked.

Incorrect Answers

A:

Batch inference is designed for asynchronous processing of submitted input collections when the selected model and API support it; it may be used to execute offline inference on a large queued dataset rather than respond to each request interactively in a production inference rollout, but the current requirement is instead to run a supported AI capability close to data in a controlled container host. That separate capability does not prove that the workload runs on the chosen container host with the required metering setup; It is unsuitable for a synchronous low-latency user chat turn, so selecting that alternative would not meet the stated acceptance check.

C:

Standard online deployments are used for request-driven inference within supported model and service limits; it may be used to serve ordinary interactive inference traffic without a pre-purchased dedicated capacity reservation in a production inference rollout, but the current requirement is instead to run a supported AI capability close to data in a controlled container host. That separate capability does not prove that the workload runs on the chosen container host with the required metering setup; They do not promise reserved throughput when a workload requires predictable provisioned capacity, so selecting that alternative would not meet the stated acceptance check.

D:

Provisioned throughput allocates purchased serving capacity for compatible model deployments subject to regional and model support; it may be used to reserve predictable model processing capacity for sustained production demand in a production inference rollout, but the current requirement is instead to run a supported AI capability close to data in a controlled container host. That separate capability does not prove that the workload runs on the chosen container host with the required metering setup; It is not the same as a pay-as-you-go standard deployment, so selecting that alternative would not meet the stated acceptance check.

E:

Fine-tuning modifies model behavior through a supported training process and produces an adapted model artifact; it may be used to adapt a pretrained model to a specialized labeled example set where that model supports tuning in a production inference rollout, but the current requirement is instead to run a supported AI capability close to data in a controlled container host. That separate capability does not prove that the workload runs on the chosen container host with the required metering setup; It is not a substitute for deploying a model or retrieving fresh private documents at query time, so selecting that alternative would not meet the stated acceptance check.

Designing and Implementing a Microsoft Azure AI Solution (AI-102) Practice Tests