Practice Exams:

Microsoft AI-103 Dumps

EXAM: Developing AI Apps and Agents on Azure
Verified By Experts
Premium Bundle
$49.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 60

Question 590

During a regional deployment, a compliance specialist must accept a change for retry backoff. The issue to be addressed is: A bursty application receives 429 responses from an inference endpoint. Which concrete verification action demonstrates that the relevant behavior works?

  1. Retry instantly without limits and amplify throttling pressure
  2. Test a retrieved instruction attempting unauthorized export
  3. Stress the endpoint and inspect retry rates and eventual failures
  4. Use bounded exponential backoff with jitter and respect server guidance

Explanation

Correct Answer

C:

To verify this change, the logistics operator should stress the endpoint and inspect retry rates and eventual failures. Operators still need the implementation to use bounded exponential backoff with jitter and respect server guidance and the preventive requirement to enforce maximum retries budgets and error visibility.

Incorrect Answers

A:

In the logistics operator acceptance review, trying to retry instantly without limits and amplify throttling pressure does not test the exact retry backoff condition where a bursty application receives 429 responses from an inference endpoint; the relevant test is to stress the endpoint and inspect retry rates and eventual failures. Acceptance should instead demonstrate that the service can stress the endpoint and inspect retry rates and eventual failures, using the described failure as the test criterion.

B:

For prompt injection regression, it can be sensible to test a retrieved instruction attempting unauthorized export if a new tool integration lets retrieved snippets influence unauthorized actions. Without the actual change to stress the endpoint and inspect retry rates and eventual failures, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

The action to use bounded exponential backoff with jitter and respect server guidance can contribute to retry backoff as a implementation step; here, a bursty application receives 429 responses from an inference endpoint. In the logistics operator acceptance review, trying to use bounded exponential backoff with jitter and respect server guidance does not test the exact retry backoff condition where a bursty application receives 429 responses from an inference endpoint; the relevant test is to stress the endpoint and inspect retry rates and eventual failures.

Question 591

At a municipal services team, the product support lead is preparing a reliability exercise. The concrete requirement is: Users wait too long before seeing any response in a streaming chat app. Which design action BEST addresses the stated need?

  1. Measure first-token latency separately from total response completion
  2. Judge streaming responsiveness only by total generated text length
  3. Run streaming requests and compare first token and final token timestamps
  4. Use bounded exponential backoff with jitter and respect server guidance

Explanation

Correct Answer

A:

The proposed implementation is to measure first-token latency separately from total response completion; this matches the operational requirement that users wait too long before seeing any response in a streaming chat app. The team can run streaming requests and compare first token and final token timestamps. The team should also track time-to-first-token and total latency SLOs.

Incorrect Answers

B:

Choosing to judge streaming responsiveness only by total generated text length cannot directly replace the specified action to measure first-token latency separately from total response completion when users wait too long before seeing any response in a streaming chat app. The actual implementation must measure first-token latency separately from total response completion, because the stated service condition is that users wait too long before seeing any response in a streaming chat app.

C:

The action to run streaming requests and compare first token and final token timestamps can contribute to token stream latency as a targeted acceptance check step; here, users wait too long before seeing any response in a streaming chat app. Choosing to run streaming requests and compare first token and final token timestamps cannot directly replace the specified action to measure first-token latency separately from total response completion when users wait too long before seeing any response in a streaming chat app.

D:

Choosing to use bounded exponential backoff with jitter and respect server guidance would direct effort toward retry backoff, whose characteristic issue is that a bursty application receives 429 responses from an inference endpoint. The relevant corrective or preventive action here remains to measure first-token latency separately from total response completion, with its own validation criteria.

Question 592

After a source update at a payment services team, the team observes this failure: Users wait too long before seeing any response in a streaming chat app. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Run streaming requests and compare first token and final token timestamps
  2. Send every short tagging query to the most expensive reasoning model
  3. Measure first-token latency separately from total response completion
  4. Judge streaming responsiveness only by total generated text length

Explanation

Correct Answer

D:

This design mistake is to judge streaming responsiveness only by total generated text length; it conflicts with the actual token stream latency behavior needed when users wait too long before seeing any response in a streaming chat app. The payment services team should instead measure first-token latency separately from total response completion, followed by a check to run streaming requests and compare first token and final token timestamps.

Incorrect Answers

A:

The alternative to run streaming requests and compare first token and final token timestamps is a targeted acceptance check step rather than the faulty assumption being diagnosed. This payment services team failure-analysis alternative of run streaming requests and compare first token and final token timestamps does not diagnose the particular token stream latency defect where users wait too long before seeing any response in a streaming chat app; the keyed mistake is judge streaming responsiveness only by total generated text length.

B:

The option to send every short tagging query to the most expensive reasoning model would normally serve a different model a versus b routing requirement, where a workload has simple tagging calls and complex reasoning questions. The alternative may be useful in its own context, but it does not directly supply judge streaming responsiveness only by total generated text length for the stated decision.

C:

The alternative to measure first-token latency separately from total response completion is a implementation step rather than the faulty assumption being diagnosed. This payment services team failure-analysis alternative of measure first-token latency separately from total response completion does not diagnose the particular token stream latency defect where users wait too long before seeing any response in a streaming chat app; the keyed mistake is judge streaming responsiveness only by total generated text length.

Question 593

The transport authority is responding to a cost review. A service administrator discovers this issue: Users wait too long before seeing any response in a streaming chat app. What corrective step should the team take to restore the intended functionality most directly?

  1. Adjust buffering or deployment capacity based on measured first-token delay
  2. Invalidate stale cached entries after the price source changes
  3. Judge streaming responsiveness only by total generated text length
  4. Run streaming requests and compare first token and final token timestamps

Explanation

Correct Answer

A:

The direct repair is to adjust buffering or deployment capacity based on measured first-token delay. Once the condition is corrected, the transport authority should run streaming requests and compare first token and final token timestamps. To reduce recurrence, it should also track time-to-first-token and total latency SLOs; those steps provide different recovery and prevention functions.

Incorrect Answers

B:

This approach relates to semantic cache invalidation: an assistant repeats obsolete product prices from a cached answer. If the team chose it instead, the targeted check to run streaming requests and compare first token and final token timestamps could still fail because the underlying required action is to adjust buffering or deployment capacity based on measured first-token delay.

C:

During transport authority remediation, the proposed step to judge streaming responsiveness only by total generated text length is not the direct restoration action for token stream latency when users wait too long before seeing any response in a streaming chat app; the keyed repair is adjust buffering or deployment capacity based on measured first-token delay. After remediation the team must run streaming requests and compare first token and final token timestamps; the action in this option does not itself establish that target recovery outcome.

D:

The action to run streaming requests and compare first token and final token timestamps can contribute to token stream latency as a targeted acceptance check step; here, users wait too long before seeing any response in a streaming chat app. During transport authority remediation, the proposed step to run streaming requests and compare first token and final token timestamps is not the direct restoration action for token stream latency when users wait too long before seeing any response in a streaming chat app; the keyed repair is adjust buffering or deployment capacity based on measured first-token delay.

Question 594

A cloud developer at a large online retailer wants the following issue to be prevented in later deployments: Users wait too long before seeing any response in a streaming chat app. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Run streaming requests and compare first token and final token timestamps
  2. Define explicit improvement criteria and review budget
  3. Adjust buffering or deployment capacity based on measured first-token delay
  4. Track time-to-first-token and total latency SLOs
  5. Judge streaming responsiveness only by total generated text length

Explanation

Correct Answer

D:

Standardizing the boundary to track time-to-first-token and total latency SLOs is preventive because the same token stream latency failure could otherwise return: users wait too long before seeing any response in a streaming chat app. For this large online retailer, the service should also measure first-token latency separately from total response completion, and tests should run streaming requests and compare first token and final token timestamps so that the ongoing requirement is measurable.

Incorrect Answers

A:

The action to run streaming requests and compare first token and final token timestamps can contribute to token stream latency as a targeted acceptance check step; here, users wait too long before seeing any response in a streaming chat app. For prevention at the large online retailer, the measure to run streaming requests and compare first token and final token timestamps is weaker or aimed at another issue than the token stream latency guardrail to track time-to-first-token and total latency SLOs needed when users wait too long before seeing any response in a streaming chat app.

B:

For reflection loop budget, it can be sensible to define explicit improvement criteria and review budget if a model repeatedly critiques its own answer without improving it. Without the actual change to track time-to-first-token and total latency SLOs, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

C:

The action to adjust buffering or deployment capacity based on measured first-token delay can contribute to token stream latency as a corrective remediation step; here, users wait too long before seeing any response in a streaming chat app. For prevention at the large online retailer, the measure to adjust buffering or deployment capacity based on measured first-token delay is weaker or aimed at another issue than the token stream latency guardrail to track time-to-first-token and total latency SLOs needed when users wait too long before seeing any response in a streaming chat app.

E:

For prevention at the large online retailer, the measure to judge streaming responsiveness only by total generated text length is weaker or aimed at another issue than the token stream latency guardrail to track time-to-first-token and total latency SLOs needed when users wait too long before seeing any response in a streaming chat app. The lasting guardrail for this specific threat is to track time-to-first-token and total latency SLOs, not simply to complete an adjacent service task.

Question 595

During a source update, a site reliability engineer must accept a change for token stream latency. The issue to be addressed is: Users wait too long before seeing any response in a streaming chat app. Which concrete verification action demonstrates that the relevant behavior works?

  1. Measure first-token latency separately from total response completion
  2. Run streaming requests and compare first token and final token timestamps
  3. Adjust buffering or deployment capacity based on measured first-token delay
  4. Measure answer groundedness with a smaller context budget
  5. Judge streaming responsiveness only by total generated text length
  6. Track time-to-first-token and total latency SLOs

Explanation

Correct Answer

B:

The targeted acceptance test is to run streaming requests and compare first token and final token timestamps. The design should first measure first-token latency separately from total response completion. A robust public library network deployment also needs to track time-to-first-token and total latency SLOs; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

The action to measure first-token latency separately from total response completion can contribute to token stream latency as a implementation step; here, users wait too long before seeing any response in a streaming chat app. In the public library network acceptance review, trying to measure first-token latency separately from total response completion does not test the exact token stream latency condition where users wait too long before seeing any response in a streaming chat app; the relevant test is to run streaming requests and compare first token and final token timestamps.

C:

The action to adjust buffering or deployment capacity based on measured first-token delay can contribute to token stream latency as a corrective remediation step; here, users wait too long before seeing any response in a streaming chat app. In the public library network acceptance review, trying to adjust buffering or deployment capacity based on measured first-token delay does not test the exact token stream latency condition where users wait too long before seeing any response in a streaming chat app; the relevant test is to run streaming requests and compare first token and final token timestamps.

D:

The option to measure answer groundedness with a smaller context budget would normally serve a different retrieved context trimming requirement, where a RAG application sends too many irrelevant passages to the model. The alternative may be useful in its own context, but it does not directly supply run streaming requests and compare first token and final token timestamps for the stated decision.

E:

In the public library network acceptance review, trying to judge streaming responsiveness only by total generated text length does not test the exact token stream latency condition where users wait too long before seeing any response in a streaming chat app; the relevant test is to run streaming requests and compare first token and final token timestamps. Acceptance should instead demonstrate that the service can run streaming requests and compare first token and final token timestamps, using the described failure as the test criterion.

F:

The action to track time-to-first-token and total latency SLOs can contribute to token stream latency as a ongoing preventive rule step; here, users wait too long before seeing any response in a streaming chat app. In the public library network acceptance review, trying to track time-to-first-token and total latency SLOs does not test the exact token stream latency condition where users wait too long before seeing any response in a streaming chat app; the relevant test is to run streaming requests and compare first token and final token timestamps.

Question 596

At a logistics operator, the security engineer is preparing a security tabletop exercise. The concrete requirement is: An agent waits sequentially for independent lookup tools. Which design action BEST addresses the stated need?

  1. Parallelize dependent write operations without ordering or authorization checks
  2. Compare end-to-end latency for sequential and concurrent read cases
  3. Rerank and limit source chunks based on relevance and evidence coverage
  4. Parallelize independent safe read-only tool calls with bounded concurrency

Explanation

Correct Answer

D:

Choosing to parallelize independent safe read-only tool calls with bounded concurrency directly targets the described parallel tool retrieval gap: an agent waits sequentially for independent lookup tools. As a lasting boundary, they also need to limit parallel calls and test cancellation across branches, because an implementation by itself does not prove safe operation.

Incorrect Answers

A:

Choosing to parallelize dependent write operations without ordering or authorization checks cannot directly replace the specified action to parallelize independent safe read-only tool calls with bounded concurrency when an agent waits sequentially for independent lookup tools. The actual implementation must parallelize independent safe read-only tool calls with bounded concurrency, because the stated service condition is that an agent waits sequentially for independent lookup tools.

B:

The action to compare end-to-end latency for sequential and concurrent read cases can contribute to parallel tool retrieval as a targeted acceptance check step; here, an agent waits sequentially for independent lookup tools. Choosing to compare end-to-end latency for sequential and concurrent read cases cannot directly replace the specified action to parallelize independent safe read-only tool calls with bounded concurrency when an agent waits sequentially for independent lookup tools.

C:

Choosing to rerank and limit source chunks based on relevance and evidence coverage would direct effort toward retrieved context trimming, whose characteristic issue is that a RAG application sends too many irrelevant passages to the model. The relevant corrective or preventive action here remains to parallelize independent safe read-only tool calls with bounded concurrency, with its own validation criteria.

Question 597

After a support escalation at a energy cooperative, the team observes this failure: An agent waits sequentially for independent lookup tools. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Parallelize dependent write operations without ordering or authorization checks
  2. Parallelize independent safe read-only tool calls with bounded concurrency
  3. Compare end-to-end latency for sequential and concurrent read cases
  4. Increase maximum output indefinitely to improve responsiveness

Explanation

Correct Answer

A:

The faulty assumption is parallelize dependent write operations without ordering or authorization checks. A functioning alternative would be to parallelize independent safe read-only tool calls with bounded concurrency, and the relevant validation would be to compare end-to-end latency for sequential and concurrent read cases.

Incorrect Answers

B:

The alternative to parallelize independent safe read-only tool calls with bounded concurrency is a implementation step rather than the faulty assumption being diagnosed. This energy cooperative failure-analysis alternative of parallelize independent safe read-only tool calls with bounded concurrency does not diagnose the particular parallel tool retrieval defect where an agent waits sequentially for independent lookup tools; the keyed mistake is parallelize dependent write operations without ordering or authorization checks.

C:

The alternative to compare end-to-end latency for sequential and concurrent read cases is a targeted acceptance check step rather than the faulty assumption being diagnosed. This energy cooperative failure-analysis alternative of compare end-to-end latency for sequential and concurrent read cases does not diagnose the particular parallel tool retrieval defect where an agent waits sequentially for independent lookup tools; the keyed mistake is parallelize dependent write operations without ordering or authorization checks.

D:

The option to increase maximum output indefinitely to improve responsiveness would normally serve a different latency under long outputs requirement, where a support bot's response time increases with unbounded verbose generation. The alternative may be useful in its own context, but it does not directly supply parallelize dependent write operations without ordering or authorization checks for the stated decision.

Question 598

The industrial maintenance team is responding to a service audit. A responsible-AI reviewer discovers this issue: An agent waits sequentially for independent lookup tools. What corrective step should the team take to restore the intended functionality most directly?

  1. Move simple cases to a validated small model and retain complex fallback
  2. Compare end-to-end latency for sequential and concurrent read cases
  3. Parallelize dependent write operations without ordering or authorization checks
  4. Rerun independent reads concurrently and preserve required join semantics

Explanation

Correct Answer

D:

The appropriate remediation is to rerun independent reads concurrently and preserve required join semantics, because the failure concerns parallel tool retrieval: an agent waits sequentially for independent lookup tools. A meaningful verification should compare end-to-end latency for sequential and concurrent read cases, and the durable operating rule should limit parallel calls and test cancellation across branches after the immediate fix.

Incorrect Answers

A:

This approach relates to model a versus b routing: a workload has simple tagging calls and complex reasoning questions. If the team chose it instead, the targeted check to compare end-to-end latency for sequential and concurrent read cases could still fail because the underlying required action is to rerun independent reads concurrently and preserve required join semantics.

B:

The action to compare end-to-end latency for sequential and concurrent read cases can contribute to parallel tool retrieval as a targeted acceptance check step; here, an agent waits sequentially for independent lookup tools. During industrial maintenance team remediation, the proposed step to compare end-to-end latency for sequential and concurrent read cases is not the direct restoration action for parallel tool retrieval when an agent waits sequentially for independent lookup tools; the keyed repair is rerun independent reads concurrently and preserve required join semantics.

C:

During industrial maintenance team remediation, the proposed step to parallelize dependent write operations without ordering or authorization checks is not the direct restoration action for parallel tool retrieval when an agent waits sequentially for independent lookup tools; the keyed repair is rerun independent reads concurrently and preserve required join semantics. After remediation the team must compare end-to-end latency for sequential and concurrent read cases; the action in this option does not itself establish that target recovery outcome.

Question 599

A solution architect at a research university wants the following issue to be prevented in later deployments: An agent waits sequentially for independent lookup tools. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Parallelize dependent write operations without ordering or authorization checks
  2. Set trace retention and sensitive data minimization requirements
  3. Limit parallel calls and test cancellation across branches
  4. Rerun independent reads concurrently and preserve required join semantics

Explanation

Correct Answer

C:

The sustained prevention measure is to limit parallel calls and test cancellation across branches. The implementation can still use the separate action to parallelize independent safe read-only tool calls with bounded concurrency, but a guardrail constrains future changes. The research university can measure whether the rule works by arranging to compare end-to-end latency for sequential and concurrent read cases.

Incorrect Answers

A:

For prevention at the research university, the measure to parallelize dependent write operations without ordering or authorization checks is weaker or aimed at another issue than the parallel tool retrieval guardrail to limit parallel calls and test cancellation across branches needed when an agent waits sequentially for independent lookup tools. The lasting guardrail for this specific threat is to limit parallel calls and test cancellation across branches, not simply to complete an adjacent service task.

B:

The action to set trace retention and sensitive data minimization requirements addresses production trace correlation rather than the decisive parallel tool retrieval condition. The research university should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to limit parallel calls and test cancellation across branches.

D:

The action to rerun independent reads concurrently and preserve required join semantics can contribute to parallel tool retrieval as a corrective remediation step; here, an agent waits sequentially for independent lookup tools. For prevention at the research university, the measure to rerun independent reads concurrently and preserve required join semantics is weaker or aimed at another issue than the parallel tool retrieval guardrail to limit parallel calls and test cancellation across branches needed when an agent waits sequentially for independent lookup tools.

Question 600

During a support escalation, a machine-learning developer must accept a change for parallel tool retrieval. The issue to be addressed is: An agent waits sequentially for independent lookup tools. Which concrete verification action demonstrates that the relevant behavior works?

  1. Stress the endpoint and inspect retry rates and eventual failures
  2. Compare end-to-end latency for sequential and concurrent read cases
  3. Parallelize dependent write operations without ordering or authorization checks
  4. Parallelize independent safe read-only tool calls with bounded concurrency

Explanation

Correct Answer

B:

To verify this change, the education technology company should compare end-to-end latency for sequential and concurrent read cases. Operators still need the implementation to parallelize independent safe read-only tool calls with bounded concurrency and the preventive requirement to limit parallel calls and test cancellation across branches.

Incorrect Answers

A:

For retry backoff, it can be sensible to stress the endpoint and inspect retry rates and eventual failures if a bursty application receives 429 responses from an inference endpoint. Without the actual change to compare end-to-end latency for sequential and concurrent read cases, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

C:

In the education technology company acceptance review, trying to parallelize dependent write operations without ordering or authorization checks does not test the exact parallel tool retrieval condition where an agent waits sequentially for independent lookup tools; the relevant test is to compare end-to-end latency for sequential and concurrent read cases. Acceptance should instead demonstrate that the service can compare end-to-end latency for sequential and concurrent read cases, using the described failure as the test criterion.

D:

The action to parallelize independent safe read-only tool calls with bounded concurrency can contribute to parallel tool retrieval as a implementation step; here, an agent waits sequentially for independent lookup tools. In the education technology company acceptance review, trying to parallelize independent safe read-only tool calls with bounded concurrency does not test the exact parallel tool retrieval condition where an agent waits sequentially for independent lookup tools; the relevant test is to compare end-to-end latency for sequential and concurrent read cases.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests