Microsoft AI-300 Online Practice Questions and Exam Preparation

Microsoft AI-300 Online Questions & Answers

• Question 71:

  You need to standardize how Fabrikam Inc. manages machine learning assets.

  Which action should you perform first?

  A. Register assets in the Azure Machine Learning registry.
  B. Create a shared Azure Machine Learning workspace.
  C. Deploy a managed online endpoint.
  D. Create a new Microsoft Foundry project.
  B. Create a shared Azure Machine Learning workspace.

  ---

  Explanation

  A shared Azure Machine Learning workspace is the first standardization boundary because it centralizes compute, datastores, jobs, models, and other project assets. A registry is useful after assets exist and need cross-workspace sharing, but it is not the first place to manage project execution. Other choices focus on deployment or Foundry rather than the workspace-level MLOps requirement.

• Question 72:

  HOTSPOT

  You manage a Retrieval-Augmented Generation (RAG) system that retrieves internal policy documents from a vector index.

  Recent analysis shows that:

  1. Retrieved results frequently include duplicated content from the same document.

  2. Retrieved chunks sometimes span unrelated policy sections.

  You review the following retrieval and ingestion configurations:

  

  You need to reduce duplicated retrieval results and improve chunk relevance across policy sections.

  For each of the following statements, select Yes if the statement is true. Otherwise, select No.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  Large chunk overlap can duplicate content across retrieved chunks, while a high similarity threshold enforces stricter filtering and can exclude relevant content. The operational rule is to tune chunking, top-k, thresholds, and retrieval strategy together because each setting can trade recall, precision, and context quality.

• Question 73:

  You need to recommend an experiment-tracking strategy that ensures consistent experiment results.

  What should you recommend?

  A. Azure Machine Learning job output logs
  B. MLflow experiment tracking
  C. Application Insights logs
  D. Azure Monitor alerts
  B. MLflow experiment tracking

  ---

  Explanation

  MLflow experiment tracking is the correct strategy for consistent experiment results because it records run parameters, metrics, artifacts, and model outputs in a structured experiment history. Job output logs and Application Insights logs are useful for diagnostics, while Azure Monitor alerts handle threshold notifications. Those other services do not replace experiment tracking.

• Question 74:

  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.

  Which file format should you use?

  A. CSV
  B. TSV
  C. JSONL
  D. JSON
  C. JSONL

  ---

  Explanation

  Fine-tuning chat models requires training data in chat-style JSONL records that represent the conversational messages and expected assistant output. Plain CSV, unstructured text, or unrelated data formats do not preserve the message roles required for supervised tuning. The other alternatives miss the structure needed to train a model on prompt and response behavior.

• Question 75:

  DRAG DROP

  A company is standardizing generative AI development across multiple teams.

  Each team requires an isolated workspace. Governance and shared connections must be centrally managed.

  You need to implement a Microsoft Foundry environment structure that supports centralized governance and team isolation.

  Which type of configuration should you use for each requirement? To answer, move the appropriate configurations to the correct requirements. You may use each configuration 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 image maps team isolation to a Microsoft Foundry project per team and shared governance or connections to a shared Foundry resource. Projects isolate team workspaces, while the shared resource centralizes platform-level configuration. Other combinations either over-share team work or fragment governance, so they do not match both requirements.

• Question 76:

  A team is developing a generative AI assistant. The team is experimenting with multiple prompt variants to improve the user experience.

  When comparing prompt variants, the team plans to assess whether the generated responses are grammatically correct.

  You need to evaluate the quality of the language from the generated responses.

  Which evaluator should you use?

  A. Coherence
  B. Textual similarity
  C. Grounded ness
  D. Fluency
  D. Fluency

  ---

  Explanation

  Comparing prompt variants with evaluation metrics is the right way to determine which prompt improves user experience before release. A deployment or storage change alone does not measure prompt quality.

  Other approaches may change the application, but they do not provide evidence that one prompt variant performs better against the requirement.

• Question 77:

  An organization is deploying several generative AI workloads by using Microsoft Foundry. Each workload must meet different requirements related to data governance, task specialization, and operational cost control.

  The organization requires models that meet the following requirements:

  1. Model behavior aligns with the task being performed.

  2. Data handling aligns with internal governance policies.

  3. Operational complexity and cost are justified by workload needs.

  You need to select the foundation model options that meet the requirements.

  Which three models can you select? (Choose three.)

  A. A model that is optimized for conversational reasoning when deploying an interactive assistant
  B. The largest available model to simplify operational management
  C. The smallest available model to minimize the usage cost
  D. A model that supports multiple input types when workloads require combined text and image analysis
  E. A model that offers enterprise governance controls when workloads process regulated business data
  B. The largest available model to simplify operational management
  C. The smallest available model to minimize the usage cost
  E. A model that offers enterprise governance controls when workloads process regulated business data

  ---

  Explanation

  Different generative AI workloads require model choices and deployment configurations that match data governance, task specialization, and performance needs. Appropriate model selection, deployment strategy, and provisioned capacity are the operational levers. Other options that ignore governance or workload characteristics would not satisfy production GenAIOps requirements across multiple workloads.

• Question 78:

  DRAG DROP

  An organization uses Microsoft Foundry to develop generative AI projects that access shared Azure resources such as storage accounts and vector databases.

  The organization s security policy requires eliminating secret key-based authentication and enforcing least-privilege access.

  You must configure identity and access so that:

  1. Services authenticate without stored credentials.

  2. Permissions are scoped appropriately across projects and shared resources.

  You need to configure the appropriate identity or access mechanism for each requirement.

  What should you configure in Microsoft Foundry to meet each requirement? To answer, move the appropriate configuration mechanisms to the correct requirements. You may use each configuration mechanism 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 points to managed identity for keyless access from Foundry workloads to Azure resources and RBAC for controlling permissions. This satisfies the requirement without embedding secrets. Other mechanisms such as service principals or keys can work in some designs, but they add credential management overhead and do not match the keyless access requirement as directly.

• Question 79:

  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 manage an Azure Machine Learning workspace. The Python script named script.py reads an argument named training_data. The training_data argument specifies the path to the training data in a file named dataset1.csv.

  You plan to run the script.py Python script as a command job that trains a machine learning model. You need to provide the command to pass the path for the dataset as a parameter value when you submit the script as a training job.

  Solution: python script.py dataset1.

  csv Does the solution meet the goal?

  A. Yes
  B. No
  B. No

  ---

  Explanation

  The proposed solution does not meet the goal if it relies on a monitoring or logging mechanism that is not experiment tracking. Training consistency requires MLflow-style tracking of parameters, metrics, and artifacts for each run. Other operational logs may help troubleshoot applications, but they are not the authoritative experiment record for this requirement.

• Question 80:

  A platform team must let multiple Machine Learning workspaces reuse approved environments and components without copying files between workspaces.

  Which capability should the team use?

  A. An Azure Machine Learning registry
  B. A managed online endpoint
  C. A local Conda environment
  D. A single shared notebook
  A. An Azure Machine Learning registry

  ---

  Explanation

  An Azure Machine Learning registry is used to share approved assets such as environments and components across workspaces. That avoids copying files and preserves versioned reuse. Online endpoints serve models, local Conda environments stay outside governed sharing, and notebooks do not provide an enterprise asset distribution mechanism.