最新のAI-300合格保証付き試験問題集の認定サンプル問題
最新AI-300テスト材料には有効なAI-300テストエンジン
質問 # 70
An Azure Machine Learning workspace processes sensitive training data.
The workspace must NOT be accessible from the public internet.
You need to restrict network access.
Which configuration should you implement?
- A. Service endpoints
- B. Private endpoints
- C. Network security groups
- D. Azure Firewall rules
正解：B
質問 # 71
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. Split traffic between deployments.
- B. Route all traffic to the new deployment.
- C. Replace the existing endpoint.
- D. Deploy the model to a batch endpoint.
正解：A
解説：
To introduce a new model version in Azure Machine Learning without service interruption, you should use Blue/Green Deployment with Traffic Splitting.
This strategy allows you to run two versions of a model simultaneously under a single Online Endpoint, gradually shifting users to the new version once it is validated.
Key Benefits
Zero Downtime: The endpoint URL stays the same; only the backend routing changes.
Easy Rollback: If the new model fails, you can instantly flip traffic back to 100% on the old version.
Risk Mitigation: Only a small subset of users is exposed to the unproven model initially.
Implementation Steps
1. Create the "Green" Deployment
Deploy the new model version as a second deployment under the existing online endpoint.
Initially, set its traffic allocation to 0%.
2. Canary Testing (Initial Split)
Shift a small percentage of traffic (e.g., 10%) to the new deployment. Monitor performance metrics, error rates, and model accuracy in a real-world environment.
3. Validation & Monitoring
Use Azure Monitor and Application Insights to compare the two deployments. Check for:
Latency: Is the new model slower?
HTTP Status Codes: Are there 4xx or 5xx errors?
Model Drift: Is the prediction quality as expected?
4. Full Rollout
If the new version is stable, increase the traffic split (e.g., 50/50) until the new model handles
100% of the traffic.
5. Cleanup
Once the "Green" deployment is confirmed as the new production standard, you can delete the old ("Blue") deployment to save costs.
Reference:
https://learn.microsoft.com/en-us/azure/well-architected/ai/operations
質問 # 72
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 a Retrieval-Augmented Generation (RAG) application built on Microsoft Foundry.
The application retrieves documents from an indexed knowledge base and generates answers for internal users.
Recent feedback indicates that answers are fluent but sometimes include information that is not supported by the documents that were retrieved.
You need to evaluate whether a proposed change improves RAG answer quality by using supported and measurable techniques.
Solution: Compare embedding vector dimensions used by the retrieval pipeline before and after the change.
Does the solution meet the goal?
- A. Yes
- B. No
正解：B
解説：
Correct:
* Review user feedback comments collected after deployment to determine whether answers appear more accurate.
Relying solely on user feedback comments is not the proper immediate action for evaluating this specific issue but it is still the best choice.
Evaluate whether your proposed change improves the Retrieval-Augmented Generation (RAG) system by reducing hallucinations (unsupported information), you need to measure faithfulness and context relevance using automated, quantifiable metrics.
The proper course of action is to implement an LLM-as-a-Judge framework using an open-source evaluation library like Ragas or TruLens.
Recommended Evaluation Plan
Establish a baseline: Run your current RAG pipeline through a test dataset of 50-100 representative user queries.
Capture the outputs: Save the user query, the exact retrieved document snippets, and the generated response for every test.
Apply the change: Deploy your proposed modification (e.g., altered prompt, different temperature, or re-ranking algorithm).Run the evaluation: Pass the test dataset through the updated pipeline to generate a new set of responses.
Compare the metrics: Use the framework to score both sets of data and mathematically verify if the change reduced unsupported claims.
Incorrect:
* Compare embedding vector dimensions used by the retrieval pipeline before and after the change.
Comparing embedding vector dimensions is not a valid or effective method for measuring RAG answer quality. Vector dimensions (e.g., 1536 or 3072) are static architectural properties of your embedding model. They do not reflect factual accuracy, semantic grounding, or the rate of hallucinations in your text generation.
* Measure token throughput and average response latency before and after applying the proposed change.
Measuring token throughput and latency is not the correct action to solve this specific problem.
Reference:
https://www.getmaxim.ai/articles/how-to-evaluate-your-rag-system/
質問 # 73
You create a binary classification model. You use the Fairlearn package to assess model fairness.
You must eliminate the need to retrain the model.
You need to implement the Fairlearn package.
Which algorithm should you use?
- A. fairiearn.reductions.ExponentiatedGradient
- B. fairlearn.postprocessing.ThresholdOptimizer
- C. fairlearn.preprocessing.CorrelationRemover
- D. fairlearn.reductions.GridSearch
正解：B
質問 # 74
You manage an Azure Machine learning workspace. You develop a machine learning model.
You must deploy the model to use a low-priority VM with a pricing discount.
You need to deploy the model.
Which compute target should you use?
- A. Local deployment
- B. Azure Container Instances (ACI)
- C. Azure Machine Learning compute clusters
- D. Azure Kubernetes Service (AKS)
正解：C
解説：
The key concept here is low-priority (spot) VMs, which are available at a significant discount because Azure can reclaim them at any time. Azure Machine Learning compute clusters are the only target in the list that directly supports low-priority VM nodes as a cost-saving configuration. You set the minimum and maximum node counts and specify that new nodes should be provisioned as low-priority. Azure Container Instances (ACI) does not support low-priority pricing. Local deployment runs on the developer ' s machine with no Azure billing model. Azure Kubernetes Service (AKS) does support spot node pools but requires significantly more infrastructure management and is not the primary mechanism for low-priority compute in Azure Machine Learning. The exam tests whether you know that AML compute clusters are the managed way to leverage low-priority discounts inside Azure Machine Learning.
Microsoft Learn Reference Topic: Create and manage Azure Machine Learning compute clusters - Low- priority VMs
質問 # 75
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 in the review screen.
You have an Azure Machine Learning workspace. You connect to a terminal session from the Notebooks page in Azure Machine Learning studio.
You plan to add a new Jupyter kernel that will be accessible from the same terminal session.
You need to perform the task that must be completed before you can add the new kernel.
Solution: Delete the Python 3.6 - AzureML kernel.
Does the solution meet the goal?
- A. Yes
- B. No
正解：B
解説：
Correct:
* Create an environment.
Incorrect:
* Delete the Python 3.6 - AzureML kernel.
* Delete the Python 3.8 - AzureML kernel.
Note:
Before you can add a new Jupyter kernel on an Azure Machine Learning compute instance terminal, you must create a Conda environment.
Required Workflow
To officially provision and expose the new kernel to your Azure Machine Learning studio Notebooks, you need to execute the following full process from your terminal session:
Create the environment: Provision a new isolated environment (e.g., using conda create -n newenv python=3.10).
Activate the environment: Run conda activate newenv.
Install dependencies: Add the required ipykernel package using conda install ipykernel or pip install ipykernel.
Register the kernel: Bind the new environment configuration to the global Jupyter directory by running:
python -m ipykernel install --user --name newenv --display-name "My New Kernel" Reference:
https://docs.azure.cn/en-us/machine-learning/how-to-access-terminal
質問 # 76
You manage an Azure Machine Learning workspace. You build a model for which you must configure a Responsible Al dashboard.
Based on what you learn from the dashboard, you must perform the following activities:
- Determine what must be done to get a desirable outcome from the
model.
- Identify the features that have the most direct effect on your
outcome of interest.
You need to select the components to use for the Responsible Al dashboard configuration.
Which two components should you add? Each correct answer presents part of the solution.
NOTE: Each correct selection is worth one point.
- A. causal
- B. error analysis
- C. counterfactuals
- D. explanation
正解：A、C
解説：
To meet your requirements, you should configure the Responsible AI (RAI) dashboard with the Counterfactuals (What-If) and Causal Analysis components.
1. Counterfactuals (What-If)
Purpose: This component helps you figure out how to change the model's output to a target result for specific instances.
Mechanism: It generates counterfactual examples (the closest possible data points) that yield a different, desirable outcome. For example, it can answer: "What is the minimum amount this user's income needs to increase for their loan application to be approved?"
2. Causal Analysis
Purpose: This component estimates the direct causal effect of specific "treatment" features on your ultimate outcome of interest.
Mechanism: Unlike standard feature importance (which only shows correlation), causal inference answers deep prescription questions. It separates pure correlation from true causation to tell you exactly how changing a real-world policy or feature directly moves your target metric.
Reference:
https://docs.azure.cn/en-us/machine-learning/concept-responsible-ai-dashboard?view=azureml-api-2
質問 # 77
You create a workspace by using Azure Machine Learning Studio.
You must run a Python SDK v2 notebook in the workspace by using Azure Machine Learning Studio. You must preserve the current values of variables set in the notebook for the current instance.
You need to maintain the state of the notebook.
What should you do?
- A. Stop the current kernel.
- B. Change the compute.
- C. Change the current kernel.
- D. Stop the compute.
正解：A
解説：
To preserve the current values of variables set in the notebook and maintain the state for the current instance, you should stop the current kernel.
Memory State Retention: In Azure Machine Learning Studio notebooks, stopping the notebook kernel halts execution while allowing the compute instance to continue running.
Avoiding State Loss: While stopping or resetting the entire compute instance will completely clear your active in-memory variables and notebook state, cleanly stopping the kernel natively preserves your current progress and data states within that active notebook session.
Reference:
https://learn.microsoft.com/en-us/azure/machine-learning/quickstart-create-resources
質問 # 78
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
正解：B
解説：
Correct:
* python script.py --training_data ${{inputs.training_data}}
The scipt is named script.py.
For the parameter use ${{inputs.training_data}}
Incorrect:
* python script.py --training_data dataset1.csv
* python script.py dataset1.csv
* python train.py --training_data training_data
Note: Read a TabularDataset, Example
In the Input object, specify the type as AssetTypes.MLTABLE, and mode as InputOutputModes.DIRECT:
* Details omitted*
job = command(
code="./src", # Local path where the code is stored
*-> command="python train.py --inputs ${{inputs.input_data}}",
inputs=my_job_inputs,
environment="<environment_name>:<version>",
compute="cpu-cluster",
)
Reference:
https://learn.microsoft.com/en-us/azure/machine-learning/how-to-read-write-data-v2
質問 # 79
Your model requires access to external APIs using sensitive credentials during inference. You must ensure credentials are not exposed in code, logs, or environment variables. What should you implement?
- A. Store in config files
- B. Use Azure Key Vault with managed identity
- C. Encrypt credentials locally
- D. Hardcode credentials
正解：B
解説：
Azure Key Vault with managed identity ensures secure access to sensitive credentials without exposing them in code or configuration. Managed identities eliminate the need for hardcoded secrets. Other approaches, such as config files or environment variables, increase the risk of accidental exposure.
質問 # 80
A data science team plans to evaluate multiple hyperparameter values automatically while training a model in Azure Machine Learning.
The tuning process must run multiple training trials without manually modifying the training script for each run.
You need to automate hyperparameter tuning for the training job.
What should you do?
- A. Adjust hyperparameters after model deployment.
- B. Run a single training job with fixed hyperparameters.
- C. Select hyperparameters based only on default model settings.
- D. Create a tuning job that runs multiple trials with different parameter values.
正解：D
解説：
To best automate hyperparameter tuning in Azure Machine Learning, you should create a Sweep Job (in SDK v2) or a HyperDrive experiment (in SDK v1).This creates a tuning job that automatically launches multiple training trials (child runs) using your single base training script without requiring manual code modifications for each run.
Key Components to Automate the Job
To set up this job successfully using the Azure Machine Learning Python SDK v2, you will define:
Parameterized Training Script: Write your code to accept hyperparameters as command-line arguments (e.g., using Python's argparse), allowing the tuning job to pass different values to each trial.
Search Space: Define the range or specific discrete/continuous choices for the values you want to test (e.g., learning rates, batch sizes).Sampling Algorithm: Choose how Azure ML should navigate your search space. Options include Random sampling, Grid sampling (testing every possible combination), or Bayesian sampling (using previous trial results to pick the next best values).Primary Metric: Specify the performance metric your script logs (like accuracy or loss) so Azure ML knows which target to optimize.
Early Termination Policy: Optional policy (like a Bandit Policy) to automatically cancel poorly performing trials early, saving you compute time and cost.
Reference:
https://learn.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive
質問 # 81
You deploy a new model version to a managed online endpoint. You must test it with 10% traffic and automatically roll back if latency or error rate increases beyond threshold. What should you configure?
- A. Manual testing workflow
- B. Traffic splitting with monitoring alerts
- C. Separate endpoint for testing
- D. Batch endpoint validation
正解：B
解説：
Traffic splitting enables controlled rollout of a new model version by directing a percentage of requests to it. Combined with monitoring alerts, it supports automated rollback when performance degrades. Separate endpoints lack built-in traffic management and do not provide seamless or automated rollback capabilities.
質問 # 82
DRAG DROP
A team deploys a classification model to production and scores incoming customer data daily.
After several weeks, business stakeholders report unexpected changes in prediction behavior, even though the endpoint remains healthy.
You need to determine whether data drift is occurring and if it is, identify the appropriate actions.
Which action should you perform for each observed signal? To answer, move the appropriate actions to the correct observed signals. 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.
正解：
解説：
質問 # 83
A data science team plans to evaluate multiple hyperparameter values automatically while training a model in Azure Machine Learning.
The tuning process must run multiple training trials without manually modifying the training script for each run.
You need to automate hyperparameter tuning for the training job.
What should you do?
- A. Adjust hyperparameters after model deployment.
- B. Manually change hyperparameter values between training runs.
- C. Duplicate the training script for each parameter combination.
- D. Create a tuning job that runs multiple trials with different parameter values.
正解：D
解説：
Correct:
* Create a tuning job that runs multiple trials with different parameter values To best automate hyperparameter tuning in Azure Machine Learning, you should create a Sweep Job (in SDK v2) or a HyperDrive experiment (in SDK v1).
This creates a tuning job that automatically launches multiple training trials (child runs) using your single base training script without requiring manual code modifications for each run.
Key Components to Automate the Job
To set up this job successfully using the Azure Machine Learning Python SDK v2, you will define:
Parameterized Training Script: Write your code to accept hyperparameters as command-line arguments (e.g., using Python's argparse), allowing the tuning job to pass different values to each trial.
Search Space: Define the range or specific discrete/continuous choices for the values you want to test (e.g., learning rates, batch sizes).Sampling Algorithm: Choose how Azure ML should navigate your search space. Options include Random sampling, Grid sampling (testing every possible combination), or Bayesian sampling (using previous trial results to pick the next best values).Primary Metric: Specify the performance metric your script logs (like accuracy or loss) so Azure ML knows which target to optimize.
Early Termination Policy: Optional policy (like a Bandit Policy) to automatically cancel poorly performing trials early, saving you compute time and cost.
Incorrect:
* Adjust hyperparameters after model deployment.
* Duplicate the training script for each parameter combination.
* Manually change hyperparameter values between training runs.
* Run a single training job with fixed hyperparameters.
* Select hyperparameters based only on default model settings.
Reference:
https://learn.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive
質問 # 84
A company plans to deploy a foundation model in Microsoft Foundry.
The mode must support the following workloads:
A customer support workload used across multiple regions
A marketing workload that must remain within a specific region due to data residency requirements You need to select the deployment type.
Which deployment type should you use for each workload? To answer, move the appropriate deployment types to the correct requirements. You may use each deployment type 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.
正解：
解説：
Explanation:
For a customer support workload used across multiple regions, Global Standard deployment is the right choice: it routes each request to the nearest available Azure region automatically, reducing latency globally and providing the highest throughput and availability. For a marketing workload that must remain within a specific region due to data residency requirements, a Data Zone Standard or single-region deployment ensures all compute and data processing occurs within a defined geographic boundary, satisfying GDPR and local data sovereignty rules. Microsoft Foundry ' s deployment types are designed around exactly this trade-off:
Global routing for performance-critical multi-region workloads, and Data Zone or Regional isolation for data- residency-constrained workloads. Choosing the wrong deployment type can result in either compliance violations or unnecessary latency.
Microsoft Learn Reference Topic: Model deployment options in Microsoft Foundry - Global, Data Zone, and Regional deployment types
質問 # 85
You create an Azure Machine Learning workspace.
You must use the Python SDK v2 to implement an experiment from a Jupyter notebook in the workspace. The experiment must log a list of numerical metrics.
You need to implement a method to log a list of numerical metrics.
Which method should you use?
- A. mlflow.log_batch()
- B. mlflow.log_metric()
- C. mlflow.log_image()
- D. mlflow.log_artifact()
正解：B
解説：
To log a list of numerical metrics using the Azure Machine Learning Python SDK v2, you should use the mlflow.log_metric() method within a loop, or mlflow.log_metrics() to log them simultaneously as a dictionary.
Reference:
https://learn.microsoft.com/en-us/answers/questions/1456554/downloading-azureml-experiment-metrics-logged-with
質問 # 86
A team deploys a machine learning model to a managed online endpoint. The team monitors model performance and data quality metrics in production.
When monitoring thresholds are exceeded, the team requires an automated operational response that notifies downstream systems.
You need to configure the monitoring solution to meet the requirements.
Which configuration should you associate with each requirement as a first step? 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.
正解：
解説：
Explanation:
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
質問 # 87
A data science team trains a classification model that predicts loan approval outcomes.
Before registering the model, the team must ensure the following:
- Predictions must not disproportionately impact protected groups.
- Prediction errors can be evaluated across different data segments.
You need to assess whether the model meets Responsible AI expectations.
Which two approaches should you use? Each correct answer presents part of the solution.
Choose two.
NOTE: Each correct selection is worth one point.
- A. Analyze error rates across defined demographic cohorts.
- B. Analyze error rates across the global cohort.
- C. Evaluate feature importance for prediction transparency.
- D. Measure endpoint latency under load.
- E. Validate inference schema compatibility.
正解：A、C
解説：
[D]
To evaluate a trained loan classification model for Responsible AI expectations--ensuring no disproportionate impact on protected groups, evaluating error across segments, and verifying prediction transparency--you can employ SHAP (SHapley Additive exPlanations) values to assess feature importance.
This approach allows you to identify which variables (e.g., credit history, debt levels) drive the model's predictions, fostering trust and fairness.
Feature Importance for Transparency: Use SHAP (model-agnostic) or LIME (local approximations) to explain why the model approved or denied a loan. These techniques identify how each feature contributes to individual predictions.
[E]
To ensure a trained loan approval classification model meets responsible AI expectations-- specifically, that it does not disproportionately impact protected groups and that errors can be evaluated across segments--you should analyze error rates across defined demographic cohorts using Fairness-Aware Machine Learning metrics.
Reference:
https://urfpublishers.com/journal/artificial-intelligence/article/view/explainable-aiml-testing- ensuring-transparency-accountability-and-compliance
https://timvero.com/blog/ethics-in-automated-lending-can-ai-make-fair-credit-decisions
質問 # 88
You manage a Retrieval-Augmented Generation (RAG) system that retrieves internal policy documents from a vector index.
Recent analysis shows that:
Retrieved results frequently include duplicated content from the same document.
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.
正解：
解説：
Explanation:
Two distinct RAG problems require two distinct solutions. When retrieved results frequently include duplicated content from the same document, the cause is overlapping chunks receiving similarly high similarity scores. The solution is Maximum Marginal Relevance reranking, which diversifies the result set by penalizing results that are semantically similar to already-selected results, reducing redundancy. When retrieved chunks sometimes span unrelated policy sections, the cause is fixed-size character chunking that splits content without regard for semantic boundaries. The solution is semantic chunking - splitting on natural sentence or section boundaries - ensuring each chunk is semantically coherent and does not straddle unrelated content. Both problems must be addressed together: reranking alone solves duplication but not relevance; semantic chunking alone solves relevance but not duplication.
Microsoft Learn Reference Topic: Optimize RAG retrieval in Azure AI Search - Chunking strategies and MMR reranking
Topic 1, Fabrikam Inc.
Background
Fabrikam Inc. is a mid-sized healthcare analytics company that provides population health dashboards and predictive insights to regional hospital systems across the United States. Fabrikam Inc. customers rely on near real time analytics to monitor patient flow, staffing needs, and readmission risks. They use multiple traditional forecasting machine learning models for predictions. Fabrikam Inc. has an established Microsoft Azure footprint. The company uses Jupyter Notebooks that run on a local server as the primary development environment. The data science team is experiencing scalability, asset management and code management issues with the current development platform. Fabrikam Inc. plans to migrate to a cloud-based development environment to mitigate the issues.
Additionally, the company plans to implement a Retrieval-Augmented Generation (RAG)-based chat application for client support.
Leadership requires the application to be developed and deployed with a low operational risk.
Current Environment
Fabrikam Inc. operates a single Azure subscription that has the following components:
Azure Data Lake Storage Gen2 that contains de-identified clinical and operational datasets Azure AI Search indexing curated analytical documents and reference materials A small set of Python-based training scripts maintained by data scientists Azure OpenAI Service with deployed foundational models A Microsoft Foundry resource for building a RAG-based solution Evaluation data has manually defined expected responses.
The current challenges faced by the data science team include the following:
Model training jobs are run manually from notebooks.
Experiment tracking is inconsistent
Model versions are registered without standardized metadata.
Deployment is performed manually by data scientists, with limited rollback capability.
The team has no standardized evaluation process for generative AI outputs.
The environment currently allows public network access. Authentication relies on user accounts rather than managed identities.
Compute targets are manually created and shared across experiments. This has led to resource contention during peak usage.
Business Requirements
Fabrikam Inc. has the following business requirements for the modernization initiative:
Provide a conversational interface that answers analytics questions by using internal documents and datasets.
Ensure that sensitive healthcare-related data is not exposed outside the Fabrikam Inc. Azure tenant.
Enable repeatable and auditable model training and deployment processes.
Support experimentation to compare prompt strategies and fine-tuned models.
Align the model with the ranked preferences and optimize behavior for the long term.
Minimize disruption to existing analytics workloads during rollout.
Technical Requirements
To support the business goals, Fabrikam Inc. identifies these technical requirements:
Use Azure Machine Learning workspaces to centrally manage data assets, models, and environments.
Implement experiment tracking and model versioning for all training jobs.
Orchestrate training and evaluation by using pipelines rather than manually running notebooks.
Deploy traditional machine learning models with support for staged rollout and rollback.
Improve RAG-based solution output quality.
Use the existing evaluation datasets that are based on real data with input-output pairs.
Apply advanced fine-tuning techniques only when prompt engineering is insufficient Issues and Constraints Fabrikam Inc. must comply with internal security policies that require the company to restrict network access and avoid long-lived secrets. The data science team has limited Azure DevOps experience, so solutions must favor managed services and automation over custom infrastructure.
Cost predictability is important. Leadership prefers serverless or managed compute options where possible but is willing to approve dedicated compute for stable production workloads.
Problem Statement
Fabrikam Inc. must design and implement an Azure-based AI operations solution that enables reliable training, evaluation, deployment, and iteration of generative AI models. The solution must support experimentation and gradual rollout while ensuring governance, security, and operational stability. The data science and platform teams must collaborate to deliver this solution by using Azure Machine Learning and Microsoft Foundry capabilities.
質問 # 89
A team manages an Azure Machine Learning workspace and deploys a model to an endpoint.
A deployed online endpoint shows inconsistent response times during periods of high traffic.
You need to identify potential performance degradation.
Which three metrics should you monitor? Each correct answer presents part of the solution.
Choose three.
NOTE: Each correct selection is worth one point.
- A. Requests per minute
- B. Request latency
- C. Connections active
- D. Dataset size
- E. Feature count
正解：A、B、C
解説：
To locate potential performance degradation in an Azure Machine Learning online endpoint during high traffic, you should monitor these three metrics:
Requests per minute: This metric tracks the volume of incoming traffic and helps identify if spikes in load correlate with slower response times.
Connections active: This monitors the total number of concurrent TCP connections from clients, which can indicate if the endpoint is reaching its capacity limits during peak periods.
Request latency: This directly measures the time taken to respond to requests, allowing you to observe exactly when and by how much performance is degrading.
Reference:
https://oneuptime.com/blog/post/2026-02-16-how-to-deploy-a-machine-learning-model-as-a-real- time-endpoint-in-azure-machine-learning/view
質問 # 90
A data science team registers an MLflow model in Azure Machine Learning.
The model must support low latency predictions and automatically scale based on incoming request volume.
You need to deploy the model.
Which deployment option should you use?
- A. Azure OpenAI endpoint
- B. Managed online endpoint
- C. Server less endpoint
- D. Batch endpoint
正解：B
解説：
The best deployment option is an Azure Machine Learning Managed Online Endpoint.
This deployment model is specifically designed for low-latency, real-time predictions and supports native autoscaling based on incoming request volume.
Turnkey MLflow Support: You can deploy your registered MLflow model directly without writing a custom scoring script or defining a docker environment.
Low Latency: Built on high-performance infrastructure designed for real-time inference workloads.
Autoscaling: Automatically scales the number of virtual machine instances up or down based on metrics like CPU usage, memory, or request queue depth.
Traffic Splitting: Supports deploying multiple model versions under one endpoint to safely test new models using blue-green deployment strategies.
Reference:
https://www.interviews.chat/questions/azure-machine-learning-engineer
質問 # 91
Drag and Drop Question
You have an existing GitHub repository containing Azure Machine Learning project files.
You need to clone the repository to your Azure Machine Learning shared workspace file system.
Which four actions should you perform in sequence? To answer, move the appropriate actions from the list of actions to the answer area and arrange them in the correct order.
NOTE: More than one order of answer choices is correct. You will receive credit for any of the correct orders you select.
正解：
解説：
質問 # 92
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 dataset 1. 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 train.py --training_data training_data
Does the solution meet the goal?
- A. Yes
- B. No
正解：B
解説：
This solution fails for two reasons. First, the script filename is wrong: the scenario specifies script.py, but the proposed solution calls train.py. This alone disqualifies the solution. Second, the input reference syntax is incorrect. In Azure ML SDK v2 command jobs, input values are injected into the command string using a placeholder syntax with double curly braces around inputs.name. The value training_data without the placeholder is just a string literal and is not resolved to the actual file path of the input data asset. The correct command syntax uses the proper placeholder so Azure ML can resolve the registered data asset and provide its local path to the script at runtime. Both errors - wrong script name and missing placeholder syntax - make this solution non-functional.
Microsoft Learn Reference Topic: Submit training jobs as command jobs in Azure Machine Learning Python SDK v2
質問 # 93
......
AI-300サンプルには正確な更新された問題がこちら：https://www.jpntest.com/shiken/AI-300-mondaishu