McAfee-Secured Website

Exam Code: AI-300

Exam Name: Operationalizing Machine Learning and Generative AI Solutions

Certification Provider: Microsoft

Corresponding Certification: Microsoft Certified: Machine Learning Operations Engineer Associate

100% Microsoft AI-300 Guaranteed Success With Testking Training Materials

Get Real AI-300 Exam Questions & Expert Verified Answers as Experienced on the Real Exam!

213 Questions & Answers with Testing Engine

"Operationalizing Machine Learning and Generative AI Solutions Exam", also known as AI-300 exam, is a Microsoft certification exam.

Pass your tests with the always up-to-date AI-300 Exam Engine. Your AI-300 training materials keep you at the head of the pack!

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Was: $137.49
Now: $124.99

Microsoft AI-300 Practice Test 10

Question 90

A compliance team needs to know which workspace produced a model later shared through the central registry. What provenance should be retained?

  1. Publish the canonical environment asset version to a separately governed central registry.
  2. Verify the requesting identity's registry-level access permissions.
  3. Retain the earlier explicit component asset version in the historical pipeline definition.
  4. Capture the origin workspace and generating training job in registered model metadata.

Explanation

Correct Answer

D:

Origin and run lineage distinguish the producing experiment from workspaces that merely consume the same registry version.

Incorrect Answers

A:

Publish the canonical environment asset version to a separately governed central registry. It addresses another case: A workspace loses a locally published environment after cleanup, but several teams still need its reviewed version. For that case, the distinguishing justification or acceptance check is: Central registries decouple shared asset availability from any one development workspace's lifecycle and cleanup policy. The present case instead states: A compliance team needs to know which workspace produced a model later shared through the central registry. It calls for Capture the origin workspace and generating training job in registered model metadata. Showing success for the other case would not establish: Origin and run lineage distinguish the producing experiment from workspaces that merely consume the same registry version.

B:

The alternative, Verify the requesting identity's registry-level access permissions, belongs to a different technical condition: A new project cannot read a model from the organization's registry although the asset exists. There the source-specific mechanism or proof is: Registry assets have access-control requirements distinct from merely being a member of the destination workspace. The condition asked about here is: A compliance team needs to know which workspace produced a model later shared through the central registry. A suitable correction is Capture the origin workspace and generating training job in registered model metadata; the targeted justification or result is: Origin and run lineage distinguish the producing experiment from workspaces that merely consume the same registry version.

C:

Retain the earlier explicit component asset version in the historical pipeline definition. That operation would be relevant if: A registry-backed component is updated, but an existing batch pipeline must reproduce last month's scores. Its corresponding reason or verification is: Immutable historical version references preserve the exact executable component while newer versions remain separately available. However the problem in this question is: A compliance team needs to know which workspace produced a model later shared through the central registry. The requirement instead calls for Capture the origin workspace and generating training job in registered model metadata; the question-specific check is: Origin and run lineage distinguish the producing experiment from workspaces that merely consume the same registry version.

Question 91

A workspace loses a locally published environment after cleanup, but several teams still need its reviewed version. What distribution arrangement reduces this dependence?

  1. Capture the origin workspace and generating training job in registered model metadata.
  2. Require review and explicit publication of approved asset versions to the shared registry.
  3. Publish the canonical environment asset version to a separately governed central registry.
  4. Coordinate explicit pinned registry version updates through a reviewed release manifest.

Explanation

Correct Answer

C:

Central registries decouple shared asset availability from any one development workspace's lifecycle and cleanup policy.

Incorrect Answers

A:

Capture the origin workspace and generating training job in registered model metadata. This is useful for the distinct issue: A compliance team needs to know which workspace produced a model later shared through the central registry. The relevant technical rationale or evidence for that issue is: Origin and run lineage distinguish the producing experiment from workspaces that merely consume the same registry version. Here the reported issue is: A workspace loses a locally published environment after cleanup, but several teams still need its reviewed version. The relevant response is: Publish the canonical environment asset version to a separately governed central registry. The first result cannot verify this different correction: Central registries decouple shared asset availability from any one development workspace's lifecycle and cleanup policy.

B:

Require review and explicit publication of approved asset versions to the shared registry. It addresses another case: A central platform team needs to control which experimental models may enter a governed production catalog. For that case, the distinguishing justification or acceptance check is: Governance at promotion time keeps unreviewed workspace models out of the shared, production-consumable catalog. The present case instead states: A workspace loses a locally published environment after cleanup, but several teams still need its reviewed version. It calls for Publish the canonical environment asset version to a separately governed central registry. Showing success for the other case would not establish: Central registries decouple shared asset availability from any one development workspace's lifecycle and cleanup policy.

D:

The alternative, Coordinate explicit pinned registry version updates through a reviewed release manifest, belongs to a different technical condition: The platform team plans to promote a newer model version to multiple consumer workspaces. There the source-specific mechanism or proof is: A release manifest controls when each consumer advances to the new shared asset rather than assuming latest always means approved. The condition asked about here is: A workspace loses a locally published environment after cleanup, but several teams still need its reviewed version. A suitable correction is Publish the canonical environment asset version to a separately governed central registry; the targeted justification or result is: Central registries decouple shared asset availability from any one development workspace's lifecycle and cleanup policy.

Question 92

The platform team plans to promote a newer model version to multiple consumer workspaces. How can it avoid each consumer silently changing at different times?

  1. Require review and explicit publication of approved asset versions to the shared registry.
  2. Coordinate explicit pinned registry version updates through a reviewed release manifest.
  3. Retain the earlier explicit component asset version in the historical pipeline definition.
  4. Register the approved environment version in a shared Azure ML registry.

Explanation

Correct Answer

B:

A release manifest controls when each consumer advances to the new shared asset rather than assuming latest always means approved.

Incorrect Answers

A:

Require review and explicit publication of approved asset versions to the shared registry. A separate source case supports that approach when: A central platform team needs to control which experimental models may enter a governed production catalog. The supporting rationale or observable result is: Governance at promotion time keeps unreviewed workspace models out of the shared, production-consumable catalog. This question describes a different issue: The platform team plans to promote a newer model version to multiple consumer workspaces. Address it through Coordinate explicit pinned registry version updates through a reviewed release manifest, with the required justification or evidence: A release manifest controls when each consumer advances to the new shared asset rather than assuming latest always means approved.

C:

Retain the earlier explicit component asset version in the historical pipeline definition. This is useful for the distinct issue: A registry-backed component is updated, but an existing batch pipeline must reproduce last month's scores. The relevant technical rationale or evidence for that issue is: Immutable historical version references preserve the exact executable component while newer versions remain separately available. Here the reported issue is: The platform team plans to promote a newer model version to multiple consumer workspaces. The relevant response is: Coordinate explicit pinned registry version updates through a reviewed release manifest. The first result cannot verify this different correction: A release manifest controls when each consumer advances to the new shared asset rather than assuming latest always means approved.

D:

Register the approved environment version in a shared Azure ML registry. It addresses another case: A team wants all regions to use the same reusable Azure ML environment definition. Local copies differ in package pins. For that case, the distinguishing justification or acceptance check is: Shared environment versions provide a consistent reference for jobs and endpoints deployed from multiple workspace locations. The present case instead states: The platform team plans to promote a newer model version to multiple consumer workspaces. It calls for Coordinate explicit pinned registry version updates through a reviewed release manifest. Showing success for the other case would not establish: A release manifest controls when each consumer advances to the new shared asset rather than assuming latest always means approved.

Question 93

A GitHub Actions workflow deploys to Azure ML using a service principal secret stored in repository settings. Security wants to eliminate long-lived secrets. Which identity mechanism is appropriate?

  1. Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required.
  2. Use distinct federated workload identities and scopes for staging and production environments.
  3. Remove secrets from source and authenticate using identity federation or an approved secret service.
  4. Configure Microsoft Entra workload identity federation using GitHub OIDC claims.

Explanation

Correct Answer

D:

Federated credentials exchange trusted workflow identity tokens for Azure access without storing a reusable client secret in GitHub.

Incorrect Answers

A:

Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required. That operation would be relevant if: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. Its corresponding reason or verification is: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases. However the problem in this question is: A GitHub Actions workflow deploys to Azure ML using a service principal secret stored in repository settings. Security wants to eliminate long-lived secrets. The requirement instead calls for Configure Microsoft Entra workload identity federation using GitHub OIDC claims; the question-specific check is: Federated credentials exchange trusted workflow identity tokens for Azure access without storing a reusable client secret in GitHub.

B:

Use distinct federated workload identities and scopes for staging and production environments. A separate source case supports that approach when: A production deployment uses the same federated identity as experimental workflows and cannot attribute risky changes to a distinct release context. The supporting rationale or observable result is: Environment-specific workload identities improve isolation, auditability, and rollback investigation without spreading shared deployment credentials. This question describes a different issue: A GitHub Actions workflow deploys to Azure ML using a service principal secret stored in repository settings. Security wants to eliminate long-lived secrets. Address it through Configure Microsoft Entra workload identity federation using GitHub OIDC claims, with the required justification or evidence: Federated credentials exchange trusted workflow identity tokens for Azure access without storing a reusable client secret in GitHub.

C:

Remove secrets from source and authenticate using identity federation or an approved secret service. This is useful for the distinct issue: An ML operations job includes cloud credentials directly in its checked-in YAML for convenience. The relevant technical rationale or evidence for that issue is: Tracked workflow files are widely replicated and reviewed; embedding credentials creates unnecessary exposure beyond the intended deployment run. Here the reported issue is: A GitHub Actions workflow deploys to Azure ML using a service principal secret stored in repository settings. Security wants to eliminate long-lived secrets. The relevant response is: Configure Microsoft Entra workload identity federation using GitHub OIDC claims. The first result cannot verify this different correction: Federated credentials exchange trusted workflow identity tokens for Azure access without storing a reusable client secret in GitHub.

Question 94

A workflow from a public fork can request a deployment identity because the federation subject allows all branches. What change restricts impersonation?

  1. Verify issuer, audience, and subject on the federated credential against the GitHub-issued token.
  2. Use an appropriately scoped GitHub App or workflow token tied to automation.
  3. Constrain the federated credential subject and issuer to the approved repository and protected workflow context.
  4. Grant only the required workspace or resource-group actions to its deployment identity.

Explanation

Correct Answer

C:

Narrow matching conditions prevent an untrusted branch or repository event from assuming the production deployment identity.

Incorrect Answers

A:

The alternative, Verify issuer, audience, and subject on the federated credential against the GitHub-issued token, belongs to a different technical condition: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. There the source-specific mechanism or proof is: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection. The condition asked about here is: A workflow from a public fork can request a deployment identity because the federation subject allows all branches. A suitable correction is Constrain the federated credential subject and issuer to the approved repository and protected workflow context; the targeted justification or result is: Narrow matching conditions prevent an untrusted branch or repository event from assuming the production deployment identity.

B:

Use an appropriately scoped GitHub App or workflow token tied to automation. That operation would be relevant if: A pipeline uses a shared personal access token to push model configuration changes into a private repository. Its corresponding reason or verification is: Automation-owned credentials can be permission-scoped and audited without tying infrastructure changes to one employee's personal account. However the problem in this question is: A workflow from a public fork can request a deployment identity because the federation subject allows all branches. The requirement instead calls for Constrain the federated credential subject and issuer to the approved repository and protected workflow context; the question-specific check is: Narrow matching conditions prevent an untrusted branch or repository event from assuming the production deployment identity.

D:

Grant only the required workspace or resource-group actions to its deployment identity. A separate source case supports that approach when: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. The supporting rationale or observable result is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps. This question describes a different issue: A workflow from a public fork can request a deployment identity because the federation subject allows all branches. Address it through Constrain the federated credential subject and issuer to the approved repository and protected workflow context, with the required justification or evidence: Narrow matching conditions prevent an untrusted branch or repository event from assuming the production deployment identity.

Question 95

A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. What is the least-privilege change?

  1. Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required.
  2. Grant only the required workspace or resource-group actions to its deployment identity.
  3. Match the federated subject claim to the exact intended GitHub owner, repository, and reference.
  4. Verify issuer, audience, and subject on the federated credential against the GitHub-issued token.

Explanation

Correct Answer

B:

Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps.

Incorrect Answers

A:

Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required. It addresses another case: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. For that case, the distinguishing justification or acceptance check is: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases. The present case instead states: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. It calls for Grant only the required workspace or resource-group actions to its deployment identity. Showing success for the other case would not establish: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps.

C:

The alternative, Match the federated subject claim to the exact intended GitHub owner, repository, and reference, belongs to a different technical condition: A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. There the source-specific mechanism or proof is: The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant. The condition asked about here is: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. A suitable correction is Grant only the required workspace or resource-group actions to its deployment identity; the targeted justification or result is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps.

D:

Verify issuer, audience, and subject on the federated credential against the GitHub-issued token. That operation would be relevant if: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. Its corresponding reason or verification is: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection. However the problem in this question is: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. The requirement instead calls for Grant only the required workspace or resource-group actions to its deployment identity; the question-specific check is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps.

Question 96

A pipeline has correct Azure credentials but users can run it manually on unreviewed experimental branches to deploy production models. Which GitHub control should be enforced?

  1. Require an approved protected GitHub environment and reviewers before production deployment jobs.
  2. Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required.
  3. Grant only the required workspace or resource-group actions to its deployment identity.
  4. Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs.

Explanation

Correct Answer

A:

GitHub environment protection can gate sensitive workflow jobs independently from possession of a valid Azure deployment identity.

Incorrect Answers

B:

Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required. This is useful for the distinct issue: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. The relevant technical rationale or evidence for that issue is: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases. Here the reported issue is: A pipeline has correct Azure credentials but users can run it manually on unreviewed experimental branches to deploy production models. The relevant response is: Require an approved protected GitHub environment and reviewers before production deployment jobs. The first result cannot verify this different correction: GitHub environment protection can gate sensitive workflow jobs independently from possession of a valid Azure deployment identity.

C:

Grant only the required workspace or resource-group actions to its deployment identity. It addresses another case: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. For that case, the distinguishing justification or acceptance check is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps. The present case instead states: A pipeline has correct Azure credentials but users can run it manually on unreviewed experimental branches to deploy production models. It calls for Require an approved protected GitHub environment and reviewers before production deployment jobs. Showing success for the other case would not establish: GitHub environment protection can gate sensitive workflow jobs independently from possession of a valid Azure deployment identity.

D:

The alternative, Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs, belongs to a different technical condition: A malicious pull request modifies GitHub Actions YAML to print deployment secrets. There the source-specific mechanism or proof is: Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution. The condition asked about here is: A pipeline has correct Azure credentials but users can run it manually on unreviewed experimental branches to deploy production models. A suitable correction is Require an approved protected GitHub environment and reviewers before production deployment jobs; the targeted justification or result is: GitHub environment protection can gate sensitive workflow jobs independently from possession of a valid Azure deployment identity.

Question 97

A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. Which OIDC configuration should be examined?

  1. Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs.
  2. Use an appropriately scoped GitHub App or workflow token tied to automation.
  3. Verify issuer, audience, and subject on the federated credential against the GitHub-issued token.
  4. Match the federated subject claim to the exact intended GitHub owner, repository, and reference.

Explanation

Correct Answer

D:

The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant.

Incorrect Answers

A:

Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs. A separate source case supports that approach when: A malicious pull request modifies GitHub Actions YAML to print deployment secrets. The supporting rationale or observable result is: Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution. This question describes a different issue: A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. Address it through Match the federated subject claim to the exact intended GitHub owner, repository, and reference, with the required justification or evidence: The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant.

B:

Use an appropriately scoped GitHub App or workflow token tied to automation. This is useful for the distinct issue: A pipeline uses a shared personal access token to push model configuration changes into a private repository. The relevant technical rationale or evidence for that issue is: Automation-owned credentials can be permission-scoped and audited without tying infrastructure changes to one employee's personal account. Here the reported issue is: A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. The relevant response is: Match the federated subject claim to the exact intended GitHub owner, repository, and reference. The first result cannot verify this different correction: The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant.

C:

Verify issuer, audience, and subject on the federated credential against the GitHub-issued token. It addresses another case: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. For that case, the distinguishing justification or acceptance check is: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection. The present case instead states: A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. It calls for Match the federated subject claim to the exact intended GitHub owner, repository, and reference. Showing success for the other case would not establish: The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant.

Question 98

A malicious pull request modifies GitHub Actions YAML to print deployment secrets. What preventive design reduces exposure?

  1. Grant only the required workspace or resource-group actions to its deployment identity.
  2. Remove secrets from source and authenticate using identity federation or an approved secret service.
  3. Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs.
  4. Use distinct federated workload identities and scopes for staging and production environments.

Explanation

Correct Answer

C:

Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution.

Incorrect Answers

A:

Grant only the required workspace or resource-group actions to its deployment identity. That operation would be relevant if: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. Its corresponding reason or verification is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps. However the problem in this question is: A malicious pull request modifies GitHub Actions YAML to print deployment secrets. The requirement instead calls for Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs; the question-specific check is: Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution.

B:

Remove secrets from source and authenticate using identity federation or an approved secret service. A separate source case supports that approach when: An ML operations job includes cloud credentials directly in its checked-in YAML for convenience. The supporting rationale or observable result is: Tracked workflow files are widely replicated and reviewed; embedding credentials creates unnecessary exposure beyond the intended deployment run. This question describes a different issue: A malicious pull request modifies GitHub Actions YAML to print deployment secrets. Address it through Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs, with the required justification or evidence: Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution.

D:

Use distinct federated workload identities and scopes for staging and production environments. This is useful for the distinct issue: A production deployment uses the same federated identity as experimental workflows and cannot attribute risky changes to a distinct release context. The relevant technical rationale or evidence for that issue is: Environment-specific workload identities improve isolation, auditability, and rollback investigation without spreading shared deployment credentials. Here the reported issue is: A malicious pull request modifies GitHub Actions YAML to print deployment secrets. The relevant response is: Avoid granting deployment credentials to untrusted pull-request events and use least-privilege jobs. The first result cannot verify this different correction: Event-specific permissions and secretless identities reduce what unreviewed contributions can access during CI execution.

Question 99

A GitHub workflow uses the default token with broad write privileges even though its job only reads source. What settings should change?

  1. Remove secrets from source and authenticate using identity federation or an approved secret service.
  2. Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required.
  3. Grant the runtime Azure identity the required storage data-plane role on that container.
  4. Match the federated subject claim to the exact intended GitHub owner, repository, and reference.

Explanation

Correct Answer

B:

Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases.

Incorrect Answers

A:

The alternative, Remove secrets from source and authenticate using identity federation or an approved secret service, belongs to a different technical condition: An ML operations job includes cloud credentials directly in its checked-in YAML for convenience. There the source-specific mechanism or proof is: Tracked workflow files are widely replicated and reviewed; embedding credentials creates unnecessary exposure beyond the intended deployment run. The condition asked about here is: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. A suitable correction is Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required; the targeted justification or result is: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases.

C:

Grant the runtime Azure identity the required storage data-plane role on that container. That operation would be relevant if: GitHub runners authenticate successfully but cannot access artifacts from a private storage container used by the deployment. Its corresponding reason or verification is: Authentication alone is not authorization to data; Azure RBAC on storage can be required in addition to workspace roles. However the problem in this question is: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. The requirement instead calls for Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required; the question-specific check is: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases.

D:

Match the federated subject claim to the exact intended GitHub owner, repository, and reference. A separate source case supports that approach when: A CI job in one repository unexpectedly authenticates as a workload identity intended for another repository. The supporting rationale or observable result is: The subject binds the trust relationship to a specific GitHub execution context, not merely a shared Azure tenant. This question describes a different issue: A GitHub workflow uses the default token with broad write privileges even though its job only reads source. Address it through Explicitly set the GITHUB_TOKEN job permissions to read-only except where write is required, with the required justification or evidence: Job-scoped permissions reduce the actions a compromised CI step can perform against repository contents and releases.

Question 100

A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. Which configuration should be validated?

  1. Verify issuer, audience, and subject on the federated credential against the GitHub-issued token.
  2. Grant only the required workspace or resource-group actions to its deployment identity.
  3. Require an approved protected GitHub environment and reviewers before production deployment jobs.
  4. Remove secrets from source and authenticate using identity federation or an approved secret service.

Explanation

Correct Answer

A:

OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection.

Incorrect Answers

B:

Grant only the required workspace or resource-group actions to its deployment identity. That operation would be relevant if: A GitHub pipeline may update one Azure ML workspace but is currently assigned subscription Owner. Its corresponding reason or verification is: Scoped RBAC limits the blast radius while retaining permissions needed for the authorized ML deployment steps. However the problem in this question is: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. The requirement instead calls for Verify issuer, audience, and subject on the federated credential against the GitHub-issued token; the question-specific check is: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection.

C:

Require an approved protected GitHub environment and reviewers before production deployment jobs. A separate source case supports that approach when: A pipeline has correct Azure credentials but users can run it manually on unreviewed experimental branches to deploy production models. The supporting rationale or observable result is: GitHub environment protection can gate sensitive workflow jobs independently from possession of a valid Azure deployment identity. This question describes a different issue: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. Address it through Verify issuer, audience, and subject on the federated credential against the GitHub-issued token, with the required justification or evidence: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection.

D:

Remove secrets from source and authenticate using identity federation or an approved secret service. This is useful for the distinct issue: An ML operations job includes cloud credentials directly in its checked-in YAML for convenience. The relevant technical rationale or evidence for that issue is: Tracked workflow files are widely replicated and reviewed; embedding credentials creates unnecessary exposure beyond the intended deployment run. Here the reported issue is: A deployment job succeeds locally but GitHub Actions cannot authenticate because the Azure identity lacks a matching audience claim. The relevant response is: Verify issuer, audience, and subject on the federated credential against the GitHub-issued token. The first result cannot verify this different correction: OIDC federation requires all trust claims to match; resource-group permission changes cannot fix token-claim rejection.

Operationalizing Machine Learning and Generative AI Solutions (AI-300) Practice Tests