Microsoft AI-300 (Operationalizing Machine Learning and Generative AI Solutions) Practice Test 4

Question 30

The security team audits Azure Machine Learning workspace access and separation of duties. An operator is tasked to remove excess permissions from a training service identity after narrowing its actual resource needs. Which mechanism provides that capability?

  1. Identity authentication check
  2. Least-privilege review
  3. Role-assignment authority
  4. Azure RBAC role assignment scope

Explanation

Correct Answer

B:

Least privilege assigns only actions and scopes needed for operation. Broad Owner-style permissions create unnecessary ability to modify unrelated resources and access policies.

Incorrect Answers

A:

Authentication establishes the caller identity and token validity; a well-scoped workspace role cannot compensate for an expired or absent credential; the outstanding need is to remove excess permissions from a training service identity after narrowing its actual resource needs.

C:

Managing Azure RBAC assignments requires specific authorization such as an appropriate access administrator role; resource contribution privileges alone should not be assumed to include access delegation; the needed capability is to remove excess permissions from a training service identity after narrowing its actual resource needs.

D:

An assignment at the workspace resource scope limits inherited permissions to that resource and any applicable children; subscription-wide grants would expand the operator's reach; the requirement, by contrast, is to remove excess permissions from a training service identity after narrowing its actual resource needs.

Question 31

A regulated Azure ML deployment must avoid embedded secrets in automated processes. The immediate design goal is to give the Azure Machine Learning workspace a nonhuman identity for supported interactions with linked resources. Which approach addresses that goal?

  1. Compute managed identity
  2. System-assigned managed identity
  3. Data-plane RBAC grant
  4. Workspace managed identity

Explanation

Correct Answer

D:

A workspace managed identity can be granted scoped access to dependencies without storing client secrets. Its permissions remain separate from those of users and deployed job identities.

Incorrect Answers

A:

Compute identities are used by workloads on that compute when appropriately configured; granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access; the requirement, by contrast, is to give the Azure Machine Learning workspace a nonhuman identity for supported interactions with linked resources.

B:

A system-assigned identity is provisioned and removed with its parent resource; it reduces independent identity management when sharing an identity across resources is not required; the team would still need to give the Azure Machine Learning workspace a nonhuman identity for supported interactions with linked resources.

C:

A valid token alone does not authorize data access; the service principal needs the appropriate blob or other data-plane permission at a scope covering the target resource; the specified outcome instead requires the team to give the Azure Machine Learning workspace a nonhuman identity for supported interactions with linked resources.

Question 32

A regulated Azure ML deployment must avoid embedded secrets in automated processes. The team is required to authorize code running on supported compute to access storage through its execution identity. Which option is the best fit?

  1. Compute managed identity
  2. Data-plane RBAC grant
  3. User-assigned managed identity
  4. System-assigned managed identity

Explanation

Correct Answer

A:

Compute identities are used by workloads on that compute when appropriately configured. Granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access.

Incorrect Answers

B:

A valid token alone does not authorize data access; the service principal needs the appropriate blob or other data-plane permission at a scope covering the target resource; the requested task remains to authorize code running on supported compute to access storage through its execution identity.

C:

A user-assigned managed identity exists as a separate Azure resource and can be attached to supported workloads; access grants follow that identity rather than any individual human account; this alternative does not achieve the requirement to authorize code running on supported compute to access storage through its execution identity.

D:

A system-assigned identity is provisioned and removed with its parent resource; it reduces independent identity management when sharing an identity across resources is not required; the actual design criterion is to authorize code running on supported compute to access storage through its execution identity.

Question 33

A regulated Azure ML deployment must avoid embedded secrets in automated processes. An operator is tasked to reuse one explicitly managed service identity across supported resources while controlling its lifecycle independently. Which mechanism provides that capability?

  1. Federated workload identity
  2. System-assigned managed identity
  3. User-assigned managed identity
  4. Compute managed identity

Explanation

Correct Answer

C:

A user-assigned managed identity exists as a separate Azure resource and can be attached to supported workloads. Access grants follow that identity rather than any individual human account.

Incorrect Answers

A:

OpenID Connect federation exchanges a trusted workflow identity assertion for Azure access; trust policies must constrain the repository, subject, audience, and intended privileges; the actual design criterion is to reuse one explicitly managed service identity across supported resources while controlling its lifecycle independently.

B:

A system-assigned identity is provisioned and removed with its parent resource; it reduces independent identity management when sharing an identity across resources is not required; the outstanding need is to reuse one explicitly managed service identity across supported resources while controlling its lifecycle independently.

D:

Compute identities are used by workloads on that compute when appropriately configured; granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access; the needed capability is to reuse one explicitly managed service identity across supported resources while controlling its lifecycle independently.

Question 34

A regulated Azure ML deployment must avoid embedded secrets in automated processes. The immediate design goal is to tie a service identity lifecycle to the Azure resource on which the identity is enabled. Which approach addresses that goal?

  1. Compute managed identity
  2. System-assigned managed identity
  3. Workspace managed identity
  4. Federated workload identity

Explanation

Correct Answer

B:

A system-assigned identity is provisioned and removed with its parent resource. It reduces independent identity management when sharing an identity across resources is not required.

Incorrect Answers

A:

Compute identities are used by workloads on that compute when appropriately configured; granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access; the needed capability is to tie a service identity lifecycle to the Azure resource on which the identity is enabled.

C:

A workspace managed identity can be granted scoped access to dependencies without storing client secrets; its permissions remain separate from those of users and deployed job identities; the team would still need to tie a service identity lifecycle to the Azure resource on which the identity is enabled.

D:

OpenID Connect federation exchanges a trusted workflow identity assertion for Azure access; trust policies must constrain the repository, subject, audience, and intended privileges; the specified outcome instead requires the team to tie a service identity lifecycle to the Azure resource on which the identity is enabled.

Question 35

A regulated Azure ML deployment must avoid embedded secrets in automated processes. The team is required to resolve a storage 403 response after a service identity successfully authenticates. Which option is the best fit?

  1. Data-plane RBAC grant
  2. Workspace managed identity
  3. Compute managed identity
  4. System-assigned managed identity
  5. User-assigned managed identity

Explanation

Correct Answer

A:

A valid token alone does not authorize data access. The service principal needs the appropriate blob or other data-plane permission at a scope covering the target resource.

Incorrect Answers

B:

A workspace managed identity can be granted scoped access to dependencies without storing client secrets; its permissions remain separate from those of users and deployed job identities; the implementation under review needs to resolve a storage 403 response after a service identity successfully authenticates.

C:

Compute identities are used by workloads on that compute when appropriately configured; granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access; the requested task remains to resolve a storage 403 response after a service identity successfully authenticates.

D:

A system-assigned identity is provisioned and removed with its parent resource; it reduces independent identity management when sharing an identity across resources is not required; this alternative does not achieve the requirement to resolve a storage 403 response after a service identity successfully authenticates.

E:

A user-assigned managed identity exists as a separate Azure resource and can be attached to supported workloads; access grants follow that identity rather than any individual human account; the actual design criterion is to resolve a storage 403 response after a service identity successfully authenticates.

Question 36

A GitHub Actions workflow starts an Azure ML training job that reads protected blobs. The workflow cannot keep a client secret, and the submitted job must authenticate using its own approved execution identity. The identity types and data-plane permissions are already assigned. Which TWO authentication integrations should the platform team use across these two trust boundaries? Choose TWO.

  1. User-assigned managed identity
  2. Compute managed identity
  3. Federated workload identity
  4. Data-plane RBAC grant
  5. System-assigned managed identity

Explanation

Correct Answers

B:

Compute identities are used by workloads on that compute when appropriately configured; granting permissions to the workspace identity alone does not necessarily grant the job's runtime identity access. Here, the actual training process presents its execution identity to Blob Storage.

C:

OpenID Connect federation exchanges a trusted workflow identity assertion for Azure access; trust policies must constrain the repository, subject, audience, and intended privileges. Here, the GitHub runner exchanges its OIDC assertion for Azure access instead of retrieving a client secret.

Incorrect Answers

A:

A user-assigned managed identity exists as a separate Azure resource and can be attached to supported workloads; access grants follow that identity rather than any individual human account; the observed fault still involves the two separate identity exchanges at the CI and training execution boundaries.

D:

A valid token alone does not authorize data access; the service principal needs the appropriate blob or other data-plane permission at a scope covering the target resource; this mechanism leaves unresolved the two separate identity exchanges at the CI and training execution boundaries.

E:

A system-assigned identity is provisioned and removed with its parent resource; it reduces independent identity management when sharing an identity across resources is not required; it does not by itself address the two separate identity exchanges at the CI and training execution boundaries.

Question 37

A cross-functional team versions different forms of training and evaluation inputs. The immediate design goal is to register a specific single file so later jobs can refer to it by asset name and version. Which approach addresses that goal?

  1. MLTable data asset
  2. Data asset lineage
  3. URI file data asset
  4. Data asset version

Explanation

Correct Answer

C:

A URI_FILE data asset represents one file location as a versioned Azure ML input. It does not promise a tabular schema or load the file into a relational table.

Incorrect Answers

A:

An MLTable asset uses an MLTable definition to identify and interpret tabular sources; it supports data-loading semantics beyond a plain directory URI; this approach leaves the team unable to register a specific single file so later jobs can refer to it by asset name and version.

B:

Job input references and execution metadata make lineage discoverable; informal file naming alone cannot reliably associate artifacts with a reproducible ML run; the needed capability is to register a specific single file so later jobs can refer to it by asset name and version.

D:

Asset versioning lets jobs bind to a specific published input snapshot or declaration; replacing the asset name without version discipline would weaken reproducibility; the team would still need to register a specific single file so later jobs can refer to it by asset name and version.

Question 38

A cross-functional team versions different forms of training and evaluation inputs. The team is required to register a collection of files as one named, versioned location for training. Which option is the best fit?

  1. Data asset lineage
  2. MLTable data asset
  3. URI folder data asset
  4. URI file data asset

Explanation

Correct Answer

C:

A URI_FOLDER asset represents a directory or prefix of data. It preserves the file-oriented input pattern rather than imposing a table schema on every member.

Incorrect Answers

A:

Job input references and execution metadata make lineage discoverable; informal file naming alone cannot reliably associate artifacts with a reproducible ML run; the team would still need to register a collection of files as one named, versioned location for training.

B:

An MLTable asset uses an MLTable definition to identify and interpret tabular sources; it supports data-loading semantics beyond a plain directory URI; the specified outcome instead requires the team to register a collection of files as one named, versioned location for training.

D:

A URI_FILE data asset represents one file location as a versioned Azure ML input; it does not promise a tabular schema or load the file into a relational table; the requested task remains to register a collection of files as one named, versioned location for training.

Question 39

A cross-functional team versions different forms of training and evaluation inputs. An operator is tasked to describe tabular data loading rules and transformations for reuse by compatible jobs. Which mechanism provides that capability?

  1. MLTable data asset
  2. Data input mode
  3. Data asset version
  4. Data asset lineage

Explanation

Correct Answer

A:

An MLTable asset uses an MLTable definition to identify and interpret tabular sources. It supports data-loading semantics beyond a plain directory URI.

Incorrect Answers

B:

Input access mode affects how bytes are exposed to job code and can influence startup and throughput; the asset's logical identity does not by itself choose every execution-time transfer behavior; this alternative does not achieve the requirement to describe tabular data loading rules and transformations for reuse by compatible jobs.

C:

Asset versioning lets jobs bind to a specific published input snapshot or declaration; replacing the asset name without version discipline would weaken reproducibility; the actual design criterion is to describe tabular data loading rules and transformations for reuse by compatible jobs.

D:

Job input references and execution metadata make lineage discoverable; informal file naming alone cannot reliably associate artifacts with a reproducible ML run; the outstanding need is to describe tabular data loading rules and transformations for reuse by compatible jobs.

Question 40

A cross-functional team versions different forms of training and evaluation inputs. The immediate design goal is to preserve a historical input reference while publishing a revised dataset under the same logical asset name. Which approach addresses that goal?

  1. Data asset version
  2. URI file data asset
  3. MLTable data asset
  4. Data input mode

Explanation

Correct Answer

A:

Asset versioning lets jobs bind to a specific published input snapshot or declaration. Replacing the asset name without version discipline would weaken reproducibility.

Incorrect Answers

B:

A URI_FILE data asset represents one file location as a versioned Azure ML input; it does not promise a tabular schema or load the file into a relational table; this approach leaves the team unable to preserve a historical input reference while publishing a revised dataset under the same logical asset name.

C:

An MLTable asset uses an MLTable definition to identify and interpret tabular sources; it supports data-loading semantics beyond a plain directory URI; the needed capability is to preserve a historical input reference while publishing a revised dataset under the same logical asset name.

D:

Input access mode affects how bytes are exposed to job code and can influence startup and throughput; the asset's logical identity does not by itself choose every execution-time transfer behavior; the requirement, by contrast, is to preserve a historical input reference while publishing a revised dataset under the same logical asset name.

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