Free Practice Questions for Microsoft AI-300 Exam

Microsoft ' s documentation on Azure Machine Learning model monitoring describes a layered alerting architecture. At the base layer, Azure ML model monitors compute drift, prediction, and data quality metrics on a scheduled basis and publish results as Azure Monitor metrics. To notify stakeholders, you create an Azure Monitor alert rule that watches these metrics and fires an action group when a threshold is breached. Action groups support email, SMS, push notifications, and webhook calls. To initiate automated retraining, the webhook call in the action group targets an Azure ML pipeline ' s REST endpoint, triggering a retraining run. Alternatively, Azure Event Grid subscriptions on AML workspace events can route model-quality events to Azure Functions that start pipelines. The separation of monitoring, alerting, notification, and remediation is intentional, allowing each component to be updated independently.

Microsoft Learn Reference Topic: Set up model monitoring for data and model quality – Azure Machine Learning model monitoring

In Azure ML command jobs, MLflow automatically creates a run context, so explicit start_run is optional. The training script must call mlflow.log_param to record each hyperparameter for the run. If the script calls this for each hyperparameter, the requirement for capturing parameters per run is met. The script must also call mlflow.log_metric to record numeric metrics such as accuracy or loss, optionally indexed by training step. Metrics and parameters logged inside an Azure ML command job are automatically associated with the correct experiment and run — no manual specification of experiment or run ID is required. If all three elements (run context, parameter logging, metric logging) are present in the script, all experiment tracking statements are True. If any element is absent, the corresponding statement is False and the tracking requirement is not met.

Microsoft Learn Reference Topic: Log metrics, parameters, and artifacts during Azure Machine Learning training runs with MLflow

For tracking changes across contributors, Git integration is the answer: by connecting the Microsoft Foundry project to a Git repository, every prompt file change is tracked as a commit with author attribution, timestamp, and diff view, and pull requests enforce review before changes reach production. The Git history provides the complete audit trail and rollback capability needed for traceability. For allowing applications to consume updated prompts without requiring redeployment, Microsoft Foundry ' s prompt management feature allows prompts to be stored and versioned as named artifacts in the project. Applications reference prompts by name and load the latest approved version at inference time, rather than having prompt text hard-coded in the application deployment artifact. This decoupling means updating a prompt is a content operation — not a code deployment — so applications automatically pick up the new prompt without any redeployment.

Microsoft Learn Reference Topic: Prompt management in Microsoft Azure AI Foundry – Git integration and dynamic prompt versioning

Correct sequence:

From the terminal window in the Azure Machine Learning interface, run the ssh-keygen command.

From the terminal window in the Azure Machine Learning interface, run the cat ~/.ssh/id_rsa.pub command.

Add a public key to the GitHub account.

From the terminal window in the Azure Machine Learning interface, run the git clone command.

Azure Machine Learning supports cloning Git repositories directly into the workspace file system from a compute instance terminal . For an SSH-based GitHub connection, the required workflow is to generate an SSH key pair, obtain the public-key value, associate that public key with the Git account, and then clone the repository using its SSH URL. Microsoft documents this exact logical sequence for Git integration with Azure Machine Learning.

First, ssh-keygen creates the private/public SSH key pair on the Azure Machine Learning compute instance. Next, the cat ~/.ssh/id_rsa.pub command displays the public-key contents so they can be copied. The public key is then added to the GitHub account, enabling GitHub to authenticate connections originating from the compute instance. The private key must remain on the compute instance and must never be uploaded to GitHub.

Finally, execute git clone with the repository ' s SSH clone URL. Azure Machine Learning documentation confirms that repositories can be cloned directly into its shared workspace file system and recommends performing Git operations from the compute-instance terminal.

Add a private key to the GitHub account is therefore the unused and incorrect action.

Study Guide Reference: Design and implement an MLOps infrastructure — source control integration, Azure Machine Learning workspace files, SSH authentication, Git repositories, and secure development workflows.

Correct sequence:

Create a data asset.

Create a Foundry Search resource.

Create a vector index.

Create a flow.

First, create a data asset so that the locally stored text files are available within the project as a managed data source. The content must then be made searchable for Retrieval-Augmented Generation.

Next, create a Foundry Search resource to provide the search infrastructure that will store and serve the indexed document content. Microsoft documents Azure AI Search as a supported vector store for Foundry and Azure Machine Learning RAG workloads.

Third, create a vector index from the source documents. During vector-index creation, the content is processed into chunks and embeddings, which enables semantic similarity retrieval. Microsoft documents that vector indexes can be created from local files, folders, or registered data assets and then consumed by an Index Lookup operation.

Finally, create a flow and configure it to query the vector index and supply the retrieved context to the connected Azure OpenAI model. Microsoft specifically documents adding an existing vector index to a prompt flow through the Index Lookup tool.

Create a connection is the unused action in this question because the scenario already specifies a connected Azure OpenAI Service model.

Study Guide Reference: Design and implement a GenAIOps infrastructure — RAG architecture, project data, vector indexing, Azure AI Search, prompt flow, and grounding with enterprise content.

The correct approach is to generate synthetic prompt-response interactions that reproduce the structure and behavioral patterns of the target task. Microsoft Foundry supports synthetic data generation specifically for cases where real production data is unavailable, scarce, sensitive, or subject to privacy and compliance restrictions. Generated examples can be converted into the structured conversational format required for fine-tuning, such as JSONL containing messages with system, user, and assistant roles.

A simulator or synthetic-data generator can construct representative conversations without exposing genuine customer interactions. This allows the team to create sufficient task-specific examples, review their quality, remove unsuitable samples, and then use the resulting dataset for supervised fine-tuning. Microsoft notes that synthetic data can preserve useful task structure while avoiding reliance on proprietary or sensitive production datasets.

Option A is incorrect because evaluation logs are primarily intended for assessment and observability and are not automatically suitable as training examples. Option C directly violates the requirement because it captures real production traffic. Option D confuses training data with observability artifacts: telemetry logs and traces describe execution behavior and diagnostics rather than producing the prompt-response examples required for fine-tuning.

Study Guide Reference: Optimize generative AI systems and model performance — synthetic data generation, supervised fine-tuning datasets, simulator-based interaction generation, privacy-preserving model customization, and training-data preparation.

Input feature distributions differ from training data: Analyze dataset drift metrics

Model accuracy drops without code changes: Review prediction and ground truth trends

Endpoint latency increases under load: Investigate scaling and infrastructure metrics

When input feature distributions differ from training data , the correct action is to analyze dataset drift metrics . Azure Machine Learning model monitoring detects data drift by comparing the statistical distributions of production model inputs against reference data, commonly the original training dataset. Supported measures include Population Stability Index, Jensen-Shannon distance, normalized Wasserstein distance, and statistical tests such as Kolmogorov-Smirnov.

When model accuracy drops without code changes , the next investigation should focus on prediction and ground-truth trends . Azure Machine Learning model-performance monitoring compares production predictions with collected actual outcomes and can calculate classification metrics such as accuracy, precision, and recall. A declining score without deployment changes may indicate concept drift, prediction drift, or changing relationships between input features and target outcomes.

When endpoint latency increases under load , the issue is operational rather than primarily statistical. The team should investigate scaling and infrastructure metrics , including request latency, requests per minute, CPU/memory utilization, throttling, and replica capacity. Microsoft recommends using endpoint metrics to determine whether compute must scale up or out.

Rebuild the inference container image is not indicated by any of the observed signals.

Study Guide Reference: Implement machine learning model lifecycle and operations — production monitoring, data drift, model-performance monitoring, endpoint observability, and scaling.