Microsoft AI-300 Practice Test 20
Question 190
As an MLOps engineer reviews a model training or inference pipeline, the following condition occurs: A churn model is registered and deployed, but online inference uses different feature lookup keys from training because the model artifact lacks the feature retrieval specification. The resolution must preserve quality gates, documented artifacts and rollback options. Which investigation best verifies the cause before changing production configuration?
- Inspect prediction and ground-truth join keys, label availability delay, inference logging and published monitoring metrics.
- Inspect registry model lifecycle status, alias assignments, endpoint references, deployment logs and retirement approvals.
- Inspect model version IDs, labels/aliases, approval evidence, endpoint deployment configuration and registry audit events.
- Inspect registered model metadata, online feature lookup mapping, feature timestamps, key names and schema contract from training.
Explanation
Correct Answer
D:Inspect registered model metadata, online feature lookup mapping, feature timestamps, key names and schema contract from training. This is the most relevant evidence for feature retrieval specification missing from registered model. Model weights alone do not specify how feature values were retrieved; production consistency requires a versioned feature contract. The matching controlled remediation is: Package and version the approved feature retrieval specification with the model and enforce the same mapping at serving time.
Incorrect Answers
A:Inspect prediction and ground-truth join keys, label availability delay, inference logging and published monitoring metrics. This would investigate accuracy monitoring lacks delayed outcome labels rather than the documented feature retrieval specification missing from registered model issue. The relevant investigation is: Inspect registered model metadata, online feature lookup mapping, feature timestamps, key names and schema contract from training.
B:Inspect registry model lifecycle status, alias assignments, endpoint references, deployment logs and retirement approvals. This would investigate retired model kept active as discoverable deployment candidate rather than the documented feature retrieval specification missing from registered model issue. The relevant investigation is: Inspect registered model metadata, online feature lookup mapping, feature timestamps, key names and schema contract from training.
C:Inspect model version IDs, labels/aliases, approval evidence, endpoint deployment configuration and registry audit events. This would investigate model registry latest alias points to unapproved experiment rather than the documented feature retrieval specification missing from registered model issue. The relevant investigation is: Inspect registered model metadata, online feature lookup mapping, feature timestamps, key names and schema contract from training.
Question 191
During a Microsoft Foundry generative AI deployment review, the following condition occurs: A customer-service agent follows imperative instructions embedded in a retrieved help-center article because the prompt mixes document contents into its high-priority system instructions. The deployment must retain governed identities, endpoint contracts and auditable release history. Which production-safe remediation should the engineer apply to this AI workflow?
- Store reviewed prompt templates and configuration in version control and deploy immutable approved revisions through the release workflow.
- Separate trusted instructions from quoted retrieved context and enforce restricted tool permissions and injection-resilient prompt handling.
- Evaluate both prompt variants against the same pinned model and representative dataset with controlled assignment and consistent metrics.
- Move the low-volume prototype to an appropriate supported consumption-based deployment or reduce provisioned capacity after review.
Explanation
Correct Answer
B:Separate trusted instructions from quoted retrieved context and enforce restricted tool permissions and injection-resilient prompt handling. This targets prompt includes untrusted retrieved instructions as system policy. Retrieved text is task data, not an authorization source; prompt composition must preserve a trust boundary around untrusted context. Verify the failure through: Examine prompt role separation, retrieval snippets, untrusted-content boundary, tool permissions and injection regression tests. Validate success with: Adversarial test documents no longer cause unauthorized agent actions while the assistant still answers legitimate questions from the content.
Incorrect Answers
A:Store reviewed prompt templates and configuration in version control and deploy immutable approved revisions through the release workflow. This could address prompt hotfix bypasses git review and loses rollback history, but it does not solve the specified prompt includes untrusted retrieved instructions as system policy failure. The matching remediation is: Separate trusted instructions from quoted retrieved context and enforce restricted tool permissions and injection-resilient prompt handling.
C:Evaluate both prompt variants against the same pinned model and representative dataset with controlled assignment and consistent metrics. This could address prompt a/b test uses different model versions in each arm, but it does not solve the specified prompt includes untrusted retrieved instructions as system policy failure. The matching remediation is: Separate trusted instructions from quoted retrieved context and enforce restricted tool permissions and injection-resilient prompt handling.
D:Move the low-volume prototype to an appropriate supported consumption-based deployment or reduce provisioned capacity after review. This could address provisioned deployment pays for idle unused capacity, but it does not solve the specified prompt includes untrusted retrieved instructions as system policy failure. The matching remediation is: Separate trusted instructions from quoted retrieved context and enforce restricted tool permissions and injection-resilient prompt handling.
Question 192
During a production Azure Machine Learning model lifecycle incident, the following condition occurs: A Python training script writes its trained pickle file only to an ephemeral worker directory, so the Azure ML submitted command job ends successfully but the model registration stage finds no artifact. The operator must safeguard training reproducibility and serving continuity. Select TWO: the strongest initial evidence and the aligned controlled correction.
- Align the published endpoint contract and client payload mapping, then deploy the reviewed scoring code and test representative requests.
- Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input.
- Add the vetted package to a versioned Azure ML environment and submit the notebook's training logic as a reproducible job.
- Inspect sweep max_concurrent_trials, compute cluster maximum nodes, GPU family quota and job scheduling telemetry.
- Write the model into the declared Azure ML job output path and track the versioned artifact used in registration.
Explanation
Correct Answers
B:Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input. This is the most relevant evidence for training job does not persist output artifacts. Managed job workers may be ephemeral; outputs must be explicitly persisted through Azure ML declared artifacts rather than private local filesystem paths. The matching controlled remediation is: Write the model into the declared Azure ML job output path and track the versioned artifact used in registration.
E:Write the model into the declared Azure ML job output path and track the versioned artifact used in registration. This targets training job does not persist output artifacts. Managed job workers may be ephemeral; outputs must be explicitly persisted through Azure ML declared artifacts rather than private local filesystem paths. Verify the failure through: Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input. Validate success with: The finished job produces a discoverable artifact that can be loaded and registered by the next pipeline stage.
Incorrect Answers
A:Align the published endpoint contract and client payload mapping, then deploy the reviewed scoring code and test representative requests. This is associated with endpoint scoring script expects wrong json schema, not the essential evidence-and-fix pair for training job does not persist output artifacts. First examine: Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input. Then correct: Write the model into the declared Azure ML job output path and track the versioned artifact used in registration.
C:Add the vetted package to a versioned Azure ML environment and submit the notebook's training logic as a reproducible job. This is associated with notebook experiment installs packages only in interactive kernel, not the essential evidence-and-fix pair for training job does not persist output artifacts. First examine: Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input. Then correct: Write the model into the declared Azure ML job output path and track the versioned artifact used in registration.
D:Inspect sweep max_concurrent_trials, compute cluster maximum nodes, GPU family quota and job scheduling telemetry. This is associated with sweep concurrency overwhelms scarce gpu quota, not the essential evidence-and-fix pair for training job does not persist output artifacts. First examine: Review job output mappings, script save path, managed output directories, job artifacts and downstream register-component input. Then correct: Write the model into the declared Azure ML job output path and track the versioned artifact used in registration.
Question 193
As an MLOps engineer reviews a model training or inference pipeline, the following condition occurs: An Azure ML online endpoint reports HTTP 400/500 errors because the deployed scoring code expects a named features field but the calling service sends a raw nested array. The resolution must preserve quality gates, documented artifacts and rollback options. Which diagnostic evidence most directly distinguishes the described AI operations failure?
- Compare application latency SLA, current endpoint type, request/response behavior, compute resources and intended inference pattern.
- Inspect request volume, concurrency, endpoint latency graphs, job timing and batch-friendly data path availability.
- Inspect request payload schema, scoring script init/run implementation, model input signature, container logs and test invocation.
- Inspect sweep max_concurrent_trials, compute cluster maximum nodes, GPU family quota and job scheduling telemetry.
Explanation
Correct Answer
C:Inspect request payload schema, scoring script init/run implementation, model input signature, container logs and test invocation. This is the most relevant evidence for endpoint scoring script expects wrong json schema. Endpoint infrastructure can be healthy while the caller and scoring script disagree on their data contract. The matching controlled remediation is: Align the published endpoint contract and client payload mapping, then deploy the reviewed scoring code and test representative requests.
Incorrect Answers
A:Compare application latency SLA, current endpoint type, request/response behavior, compute resources and intended inference pattern. This would investigate online scoring requirement deployed as batch endpoint rather than the documented endpoint scoring script expects wrong json schema issue. The relevant investigation is: Inspect request payload schema, scoring script init/run implementation, model input signature, container logs and test invocation.
B:Inspect request volume, concurrency, endpoint latency graphs, job timing and batch-friendly data path availability. This would investigate overnight bulk scoring saturates interactive online endpoint rather than the documented endpoint scoring script expects wrong json schema issue. The relevant investigation is: Inspect request payload schema, scoring script init/run implementation, model input signature, container logs and test invocation.
D:Inspect sweep max_concurrent_trials, compute cluster maximum nodes, GPU family quota and job scheduling telemetry. This would investigate sweep concurrency overwhelms scarce gpu quota rather than the documented endpoint scoring script expects wrong json schema issue. The relevant investigation is: Inspect request payload schema, scoring script init/run implementation, model input signature, container logs and test invocation.
Question 194
During a Microsoft Foundry generative AI deployment review, the following condition occurs: A Foundry application uses a supported serverless model endpoint, but only a developer's personal key works because the production managed identity lacks the required inference access. The deployment must retain governed identities, endpoint contracts and auditable release history. Which investigation best verifies the cause before changing production configuration?
- Examine billed PTU hours, token utilization, workload predictability, available pay-as-you-go endpoints and expected launch load.
- Inspect Bicep template resource references, model deployment name outputs, CI step dependencies and Azure deployment operations.
- Compare real tokenized input sizes, selected model version limits, deployment configuration and representative failure logs.
- Inspect endpoint authentication modes, deployment authorization, production managed identity, assigned roles and failed token exchanges.
Explanation
Correct Answer
D:Inspect endpoint authentication modes, deployment authorization, production managed identity, assigned roles and failed token exchanges. This is the most relevant evidence for serverless model endpoint unavailable to chosen identity. Serverless deployment mode does not remove the need for correct identity and model endpoint authorization. The matching controlled remediation is: Configure the supported identity-based inference grant and connection details for the production workload without sharing personal credentials.
Incorrect Answers
A:Examine billed PTU hours, token utilization, workload predictability, available pay-as-you-go endpoints and expected launch load. This would investigate provisioned deployment pays for idle unused capacity rather than the documented serverless model endpoint unavailable to chosen identity issue. The relevant investigation is: Inspect endpoint authentication modes, deployment authorization, production managed identity, assigned roles and failed token exchanges.
B:Inspect Bicep template resource references, model deployment name outputs, CI step dependencies and Azure deployment operations. This would investigate foundry bicep rollout omits model deployment dependencies rather than the documented serverless model endpoint unavailable to chosen identity issue. The relevant investigation is: Inspect endpoint authentication modes, deployment authorization, production managed identity, assigned roles and failed token exchanges.
C:Compare real tokenized input sizes, selected model version limits, deployment configuration and representative failure logs. This would investigate foundation model selected without required context window rather than the documented serverless model endpoint unavailable to chosen identity issue. The relevant investigation is: Inspect endpoint authentication modes, deployment authorization, production managed identity, assigned roles and failed token exchanges.
Question 195
While preparing a secure Foundry model or agent release, the following condition occurs: An agent receives a newly released model under full production traffic before anyone verifies tool calling and safety regressions. The remediation must meet enterprise access and reliability requirements without a broad bypass. Which targeted Azure AI operations change best addresses the documented problem?
- Store reviewed prompt templates and configuration in version control and deploy immutable approved revisions through the release workflow.
- Route a controlled percentage of requests to the candidate and promote only after automated quality, safety and latency gates pass.
- Build and test a compatible version-pinned serving environment and promote the corrected managed model deployment.
- Pin an approved model/deployment version where supported and pass candidate upgrades through evaluation and controlled promotion.
Explanation
Correct Answer
B:Route a controlled percentage of requests to the candidate and promote only after automated quality, safety and latency gates pass. This targets new foundation model replaces production without canary. Gradual rollout with a known-good model mitigates production blast radius when foundation model behavior changes. Verify the failure through: Inspect deployment routing weights, evaluated model versions, incident timing, canary quality dashboards and rollback path. Validate success with: Failed canary metrics restore 100% of traffic to the last approved deployment without changing unrelated agent code. This matters here because while preparing a secure Foundry model or agent release, the following condition occurs: An agent receives a newly….
Incorrect Answers
A:Store reviewed prompt templates and configuration in version control and deploy immutable approved revisions through the release workflow. This could address prompt hotfix bypasses git review and loses rollback history, but it does not solve the specified new foundation model replaces production without canary failure. The matching remediation is: Route a controlled percentage of requests to the candidate and promote only after automated quality, safety and latency gates pass.
C:Build and test a compatible version-pinned serving environment and promote the corrected managed model deployment. This could address managed foundation model deployment lacks compatible runtime image, but it does not solve the specified new foundation model replaces production without canary failure. The matching remediation is: Route a controlled percentage of requests to the candidate and promote only after automated quality, safety and latency gates pass.
D:Pin an approved model/deployment version where supported and pass candidate upgrades through evaluation and controlled promotion. This could address foundry deployment alias changes model behavior unexpectedly, but it does not solve the specified new foundation model replaces production without canary failure. The matching remediation is: Route a controlled percentage of requests to the candidate and promote only after automated quality, safety and latency gates pass.
Question 196
During a Microsoft Foundry generative AI deployment review, the following condition occurs: A custom transformer model is deployed on managed compute but inference containers crash on startup because the image's tokenizer library is incompatible with the model artifact. The deployment must retain governed identities, endpoint contracts and auditable release history. What should the engineer examine first to confirm the specific AI-300 workload issue?
- Examine model package runtime requirements, container startup traceback, tokenizer/config files and managed compute environment image digest.
- Verify region-specific model capacity, PTU quota, requested deployment units, existing reservations and allowed scaling alternatives.
- Examine prompt role separation, retrieval snippets, untrusted-content boundary, tool permissions and injection regression tests.
- Compare prompt commit hashes, foundation model version, tool JSON schema, serving configuration and prior approved release bundle.
Explanation
Correct Answer
A:Examine model package runtime requirements, container startup traceback, tokenizer/config files and managed compute environment image digest. This is the most relevant evidence for managed foundation model deployment lacks compatible runtime image. Managed inference still requires the model weights and runtime dependencies to be compatible; compute provisioning cannot fix a broken model-serving image. The matching controlled remediation is: Build and test a compatible version-pinned serving environment and promote the corrected managed model deployment.
Incorrect Answers
B:Verify region-specific model capacity, PTU quota, requested deployment units, existing reservations and allowed scaling alternatives. This would investigate provisioned model update exceeds regional capacity ceiling rather than the documented managed foundation model deployment lacks compatible runtime image issue. The relevant investigation is: Examine model package runtime requirements, container startup traceback, tokenizer/config files and managed compute environment image digest.
C:Examine prompt role separation, retrieval snippets, untrusted-content boundary, tool permissions and injection regression tests. This would investigate prompt includes untrusted retrieved instructions as system policy rather than the documented managed foundation model deployment lacks compatible runtime image issue. The relevant investigation is: Examine model package runtime requirements, container startup traceback, tokenizer/config files and managed compute environment image digest.
D:Compare prompt commit hashes, foundation model version, tool JSON schema, serving configuration and prior approved release bundle. This would investigate prompt contract rollback ignores associated model version rather than the documented managed foundation model deployment lacks compatible runtime image issue. The relevant investigation is: Examine model package runtime requirements, container startup traceback, tokenizer/config files and managed compute environment image digest.
Question 197
While optimizing a production RAG or fine-tuned model service, the following condition occurs: A domain RAG service changes the query embedding model but keeps document vectors created by a different model, producing irrelevant similarity matches. The resulting model must meet held-out quality and security criteria. Select TWO: the strongest initial evidence and the aligned controlled correction.
- Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs.
- Tune chunk size and overlap with structure-aware splitting so qualifying rules and exceptions remain retrievable together.
- Examine question-set provenance, tuning history, held-out set separation, document version and evaluator configuration.
- Use a governed retrieval-augmented grounding path for volatile factual content and reserve fine-tuning for stable behavior or output-format adaptation.
- Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model.
Explanation
Correct Answers
A:Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. This is the most relevant evidence for embedding model replaced without reindexing stored vectors. Embedding vectors from different model spaces are not generally interchangeable even when their dimension counts happen to match. The matching controlled remediation is: Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model.
E:Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model. This targets embedding model replaced without reindexing stored vectors. Embedding vectors from different model spaces are not generally interchangeable even when their dimension counts happen to match. Verify the failure through: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. Validate success with: Representative domain queries retrieve relevant passages with aligned embedding provenance and the old index can be rolled back safely.
Incorrect Answers
B:Tune chunk size and overlap with structure-aware splitting so qualifying rules and exceptions remain retrievable together. This is associated with rag retrieval chunks truncate critical policy exception, not the essential evidence-and-fix pair for embedding model replaced without reindexing stored vectors. First examine: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. Then correct: Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model.
C:Examine question-set provenance, tuning history, held-out set separation, document version and evaluator configuration. This is associated with offline rag evaluator leaks training questions into test set, not the essential evidence-and-fix pair for embedding model replaced without reindexing stored vectors. First examine: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. Then correct: Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model.
D:Use a governed retrieval-augmented grounding path for volatile factual content and reserve fine-tuning for stable behavior or output-format adaptation. This is associated with fine-tuning task selected when retrieval would better handle changing facts, not the essential evidence-and-fix pair for embedding model replaced without reindexing stored vectors. First examine: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. Then correct: Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model.
Question 198
During a regulated Azure ML workspace infrastructure review, the following condition occurs: A model platform pipeline applies Bicep changes immediately and breaks a production workspace because it never checks syntax, resource drift or the expected scope. The team must preserve approved access, model provenance and private network constraints. What approved correction resolves the failure without modifying unrelated healthy services?
- Gate production deployment behind approved GitHub environment protection, protected refs and status checks while keeping test deployments separate.
- Add Bicep validation, targeted what-if reviews and a controlled environment promotion stage before production apply.
- Rebuild or correct the versioned MLTable asset transformation so the expected columns and types are supplied to downstream training.
- Require protected pull requests and automated component tests before changes to production training code are merged.
Explanation
Correct Answer
B:Add Bicep validation, targeted what-if reviews and a controlled environment promotion stage before production apply. This targets iac workflow lacks repeatable infrastructure validation. Infrastructure automation should detect configuration drift and unexpected destructive changes before a production deployment proceeds. Verify the failure through: Inspect GitHub workflow jobs, Bicep build/what-if steps, test deployment history, parameter values and release approvals. Validate success with: Invalid or overbroad infrastructure changes fail validation and intended workspace updates deploy reproducibly after review.
Incorrect Answers
A:Gate production deployment behind approved GitHub environment protection, protected refs and status checks while keeping test deployments separate. This could address github deployment pipeline omits secure environment approvals, but it does not solve the specified iac workflow lacks repeatable infrastructure validation failure. The matching remediation is: Add Bicep validation, targeted what-if reviews and a controlled environment promotion stage before production apply.
C:Rebuild or correct the versioned MLTable asset transformation so the expected columns and types are supplied to downstream training. This could address mltable data asset schema mismatches training code, but it does not solve the specified iac workflow lacks repeatable infrastructure validation failure. The matching remediation is: Add Bicep validation, targeted what-if reviews and a controlled environment promotion stage before production apply.
D:Require protected pull requests and automated component tests before changes to production training code are merged. This could address notebook merge bypasses protected evaluation tests, but it does not solve the specified iac workflow lacks repeatable infrastructure validation failure. The matching remediation is: Add Bicep validation, targeted what-if reviews and a controlled environment promotion stage before production apply.
Question 199
While optimizing a production RAG or fine-tuned model service, the following condition occurs: A domain RAG service changes the query embedding model but keeps document vectors created by a different model, producing irrelevant similarity matches. The resulting model must meet held-out quality and security criteria. Which investigation best verifies the cause before changing production configuration?
- Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs.
- Inspect top-k search scores, threshold, query normalization, relevant passages and test question recall.
- Inspect registered fine-tuned model version, retained checkpoints, base model lineage, tokenizer files, serving configuration and retention policy.
- Inspect train/evaluation query overlap, hard negative quality, domain-term coverage, retrieval recall and candidate embedding versions.
Explanation
Correct Answer
A:Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. This is the most relevant evidence for embedding model replaced without reindexing stored vectors. Embedding vectors from different model spaces are not generally interchangeable even when their dimension counts happen to match. The matching controlled remediation is: Re-embed and reindex the document corpus with a compatible versioned embedding model, or keep query embeddings pinned to the existing index model. This matters here because while optimizing a production RAG or fine-tuned model service, the following condition occurs: A domain RAG service….
Incorrect Answers
B:Inspect top-k search scores, threshold, query normalization, relevant passages and test question recall. This would investigate rag similarity threshold filters all relevant evidence rather than the documented embedding model replaced without reindexing stored vectors issue. The relevant investigation is: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs.
C:Inspect registered fine-tuned model version, retained checkpoints, base model lineage, tokenizer files, serving configuration and retention policy. This would investigate fine-tune production endpoint references deleted training artifact rather than the documented embedding model replaced without reindexing stored vectors issue. The relevant investigation is: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. This matters here because while optimizing a production RAG or fine-tuned model service, the following condition occurs: A domain RAG service….
D:Inspect train/evaluation query overlap, hard negative quality, domain-term coverage, retrieval recall and candidate embedding versions. This would investigate specialized embedding adaptation overfits small terminology set rather than the documented embedding model replaced without reindexing stored vectors issue. The relevant investigation is: Compare query and index embedding model versions, vector dimensions, index schema, retrieval score distributions and migration logs. This matters here because while optimizing a production RAG or fine-tuned model service, the following condition occurs: A domain RAG service….
Question 200
While preparing a secure Foundry model or agent release, the following condition occurs: A global customer-support agent project is provisioned in a region that does not offer the selected foundation model deployment type, preventing its evaluation pipeline from starting. The remediation must meet enterprise access and reliability requirements without a broad bypass. What should the engineer examine first to confirm the specific AI-300 workload issue?
- Inspect runtime managed identity, search service RBAC and index permissions, connector authentication, audit failures and client requests.
- Inspect Foundry project and resource regions, model deployment regional availability, enterprise residency constraints and API error codes.
- Check application response SLA, workload concurrency, chosen Foundry model endpoint type and current request queuing behavior.
- Inspect workflow identity claims, login failure logs, secret references, federated identity options and target resource permissions.
Explanation
Correct Answer
B:Inspect Foundry project and resource regions, model deployment regional availability, enterprise residency constraints and API error codes. This is the most relevant evidence for foundry project created in mismatched resource location. A Foundry project depends on its parent resource location and model/service availability; creating a project does not guarantee every model is deployable there. The matching controlled remediation is: Deploy the governed Foundry resource and project in a supported compliant region with an available model option.
Incorrect Answers
A:Inspect runtime managed identity, search service RBAC and index permissions, connector authentication, audit failures and client requests. This would investigate agent runtime managed identity cannot read retrieval data rather than the documented foundry project created in mismatched resource location issue. The relevant investigation is: Inspect Foundry project and resource regions, model deployment regional availability, enterprise residency constraints and API error codes.
C:Check application response SLA, workload concurrency, chosen Foundry model endpoint type and current request queuing behavior. This would investigate interactive inference deployed on unsuitable batch-style compute rather than the documented foundry project created in mismatched resource location issue. The relevant investigation is: Inspect Foundry project and resource regions, model deployment regional availability, enterprise residency constraints and API error codes.
D:Inspect workflow identity claims, login failure logs, secret references, federated identity options and target resource permissions. This would investigate foundry service principal credential rotated without updating pipeline rather than the documented foundry project created in mismatched resource location issue. The relevant investigation is: Inspect Foundry project and resource regions, model deployment regional availability, enterprise residency constraints and API error codes.