Microsoft AI-300 Online Practice Questions and Exam Preparation

Microsoft AI-300 Online Questions & Answers

• Question 1:

  HOTSPOT

  You manage an Azure Machine Learning workspace named workspace1 by using the Python SDK v2.

  The default datastore of workspace1 contains a folder named sample_data. The folder structure contains the following content:

  

  You write Python SDK v2 code to materialize the data from the files in the sample_data folder into a Pandas data frame.

  You need to complete the Python SDK v2 code to use the MLTable folder as the materialization blueprint.

  How should you complete the code? To answer, select the appropriate options in the answer area.

  NOTE: Each correct selection is worth one point

  

  

  ---

  The answer uses the MLTable folder rather than an individual text file. In Azure Machine Learning, an MLTable is loaded from the folder that contains the MLTable metadata, so the path should target the sample_data directory. Other paths would either point to raw files or wildcard file matches and would not load the table definition required by the requirement.

• Question 2:

  A generative AI application has predictable high request volume. The team must reserve serving capacity to improve throughput consistency.

  Which model deployment setting should the team consider?

  A. Provisioned throughput units
  B. A data asset version
  C. A feature retrieval specification
  D. A compute cluster minimum node count for training only
  A. Provisioned throughput units

  ---

  Explanation

  Provisioned throughput units reserve model-serving capacity for high-volume generative AI workloads, improving throughput predictability. Data asset versions, feature retrieval specifications, and training cluster settings solve different Machine Learning lifecycle problems. The requirement is serving capacity for a generative AI model, so provisioned throughput is the right match.

• Question 3:

  Note: This question is part of a series of questions that present the same scenario. Each question in the series contains a unique solution that might meet the stated goals. Some question sets might have more than one correct solution, while others might not have a correct solution.

  After you answer a question in this section, you will NOT be able to return to it. As a result, these questions will not appear on the review screen.

  You work in Microsoft Foundry with a prompt flow.

  You must manually evaluate prompts and compare results across prompt variants.

  You need to capture the inputs, outputs, token usage, and latencies for each flow run for the evaluation.

  Solution: Create prompt variants and compare their outputs in the Evaluation experience.

  Does the solution meet the goal?

  A. Yes
  B. No
  B. No

  ---

  Explanation

  The proposed solution does not meet the goal when it fails to provide the required prompt versioning or review behavior. Prompt changes need source control and controlled evaluation so teams can compare variants and roll back safely.

  Other ad hoc update methods may change the application, but they do not satisfy the governance requirement in the scenario.

• Question 4:

  HOTSPOT

  A team trains an MLflow model that scores customer churn risk. The model will be consumed by different downstream systems.

  One system requests predictions synchronously during customer interactions. Another system submits files containing millions of records for scheduled scoring.

  You need to deploy the model by using managed inference options that match each usage pattern.

  Which option should you use for each usage pattern? To answer, select the appropriate options in the answer area.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  The answer pairs synchronous, low-latency prediction with an online endpoint and high-volume scheduled scoring with batch-style managed inference. Online endpoints are built for real-time request and response workloads, while batch endpoints process large files or datasets asynchronously. Other deployment choices do not satisfy both usage requirements because they either store artifacts or run training rather than serve predictions.

• Question 5:

  A team evaluates a generative AI agent before release. The team needs metrics for whether answers are supported by source content and whether responses are relevant to the user request.

  Which evaluation categories should be included?

  A. Groundedness and relevance
  B. Data drift and model archive state
  C. Compute node utilization and Git branch count
  D. Endpoint traffic percentage and datastore type
  A. Groundedness and relevance

  ---

  Explanation

  Groundedness measures whether responses are supported by source content, and relevance measures whether responses address the user request. Data drift, archive state, utilization, and traffic percentage are useful in other operations areas, but they do not evaluate generated answer quality against the requirement. The key rule is to match metrics to the quality dimension being tested.

• Question 6:

  You have a deployment of an Azure OpenAI Service base model.

  You plan to fine-tune the model.

  You need to prepare a file that contains training data for multi-turn chat.

  Which file encoding method should you use?

  A. ISO-8859-1
  B. UTF-16
  C. UTF-8
  D. ASCII
  C. UTF-8

  ---

  Explanation

  Multi-turn chat fine-tuning data must preserve the sequence of system, user, and assistant messages so the model can learn context across turns. A flat record without roles or conversation order does not satisfy the training requirement. The other formats may store text, but they do not represent the conversational structure needed for a chat model.

• Question 7:

  DRAG DROP

  A team maintains Infrastructure as Code (IaC) templates to provision Azure Machine Learning resources.

  Provisioning must be triggered by changes in the templates and executed without manual intervention.

  You need to automate resource provisioning.

  Which action should you take for each requirement? To answer, move the appropriate actions to the correct requirements. You may use each action once, more than once, or not at all. You may need to move the split bar between panes or scroll to view content.

  NOTE: Each correct selection is worth one point.

  Select and Place:

  

  

  ---

  The answer image uses GitHub Actions to trigger provisioning from repository changes and secure Azure authentication during workflow execution. That matches an infrastructure-as-code requirement because templates are applied by an automated workflow. Other manual or local commands would not provide the same repeatable, reviewed, repository-driven provisioning path.

• Question 8:

  An organization validates generative AI applications during CI/CD Microsoft Foundry.

  Evaluation must run automatically and block releases when quality thresholds are NOT met.

  Manual evaluation is no longer acceptable.

  Evaluation must use both predefined quality metrics and custom safety checks.

  You need to implement an automated evaluation workflow that supports both built-in and custom metrics.

  What should you do?

  A. Enable application tracing to collect runtime telemetry.
  B. Review evaluation results manually after deployment.
  C. Monitor latency metrics during model inference.
  D. Implement an evaluation step by using GitHub Actions.
  D. Implement an evaluation step by using GitHub Actions.

  ---

  Explanation

  Automated evaluation workflows are used in CI/CD to run quality checks and block releases when thresholds are not met. Manual review alone would not satisfy the automation requirement. Other monitoring tools can observe production behavior, but the scenario asks for release-time validation before deployment continues.

• Question 9:

  A team manages an Azure Machine Learning workspace where they deploy models to online endpoints.

  The team needs to introduce a new version of a model to production without disrupting existing users. The team must validate the new version before full rollout.

  You need to reduce risk during deployment.

  What should you do?

  A. Deploy the model to a batch endpoint.
  B. Split traffic between deployments.
  C. Replace the existing endpoint.
  D. Route all traffic to the new deployment.
  B. Split traffic between deployments.

  ---

  Explanation

  Splitting traffic between deployments is the safe rollout strategy for testing a new model version while existing users continue to use the current deployment. Replacing the endpoint or routing all traffic to the new deployment increases risk, and a batch endpoint does not serve the same real-time production pattern. The requirement is controlled validation before full rollout.

• Question 10:

  HOTSPOT

  You manage an Azure Machine Learning workspace named workspace1 by using the Python SDK v2.

  You create a General Purpose v2 Azure storage account named mlstorage1.

  The storage account includes a publicly accessible container named mlcontainer1. The container stores 10 blobs with files in the CSV format.

  You must develop Python SDK v2 code to create a data asset referencing all blobs in the container named mlcontainer1.

  You need to complete the Python SDK v2 code.

  How should you complete the code? To answer, select the appropriate options in the answer area.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  The answer uses a folder-style data asset because the storage reference points to a blob container rather than to one file. A URI folder asset is appropriate when the registered asset represents a collection of files under a container or folder path. Other asset types would imply a single file or an MLTable definition that is not shown in the image.