AI-300 Sample Questions Answers

No. Removing the existing Python 3.6 - AzureML kernel is not a prerequisite for adding a new kernel. Azure Machine Learning supports multiple installed Jupyter kernels on the same compute instance, and notebooks can select among those kernels without requiring one existing kernel to be removed first. Microsoft states that the notebook environment automatically discovers all Jupyter kernels installed on the connected compute instance.

The supported new-kernel workflow begins by creating a separate environment, for example with conda create --name newenv, and activating it. The required pip and ipykernel packages are then installed in the environment before registering it with Jupyter by using python -m ipykernel install --user --name newenv.

Deleting an existing AzureML kernel provides no technical benefit to that workflow. Microsoft also explicitly cautions users not to delete Conda environments or Jupyter kernels they did not create , because doing so can interfere with built-in Jupyter/JupyterLab functionality on the compute instance.

Therefore, the existing Python 3.6 kernel should remain untouched; the new kernel should be installed alongside it in its own environment.

Study Guide Reference: Design and implement an MLOps infrastructure — Jupyter kernel management, Conda environments, Azure Machine Learning compute instances, and development environment isolation.

Because you want to reference all blobs in a container rather than a single file, the correct asset type is AssetTypes.URI_FOLDER. This type points to a directory-level URI, allowing Azure ML to traverse all files within it. For a publicly accessible Azure Blob Storage container, the URI follows the pattern https://storage_account.blob.core.windows.net/container_name/. You then create the Data object with the correct type, path, name, and version parameters, and register it in the workspace using ml_client.data.create_or_update. Do not use AssetTypes.URI_FILE, which references a single file. Do not use AssetTypes.MLTABLE unless you have an MLTable YAML descriptor. URI_FOLDER is the correct choice for referencing a collection of CSV blobs in an Azure Blob Storage container, allowing Azure ML to discover and process all files within the specified path.

Microsoft Learn Reference Topic: Create and manage data assets in Azure Machine Learning Python SDK v2 – URI_FOLDER

Technique: Grid search

Constraint: Demographic parity

The appropriate mitigation technique is GridSearch with a DemographicParity constraint. Fairlearn provides reduction-based mitigation algorithms that retrain a standard estimator using differently weighted training data while enforcing a specified fairness constraint. GridSearch is one of Fairlearn ' s reduction algorithms for binary classification and can search across candidate models representing different trade-offs between predictive performance and fairness.

The appropriate constraint is Demographic parity because the model determines whether individuals receive an opportunity—in this case, selection and admission to a medical trial . Microsoft categorizes decisions that extend or withhold opportunities or resources as allocation harms and specifically identifies demographic parity as a parity constraint intended to mitigate allocation disparities in binary classification.

Demographic parity seeks comparable rates of positive predictions across groups defined by the sensitive attribute. Here, Ethnicity is the sensitive feature, so the objective is to reduce disparities in selection rates between ethnic groups.

False-positive-rate parity has a different purpose: it specifically requires comparable false-positive rates across groups among cases whose true label is negative. The scenario does not identify unequal false-positive rates as the problem.

An Azure AI Hub is the top-level governance container in Microsoft Foundry: it holds shared connections to Azure OpenAI, Azure AI Search, Azure Storage, and other services; it defines network isolation policies; it manages billing and quota at the organizational level. Multiple teams share these resources without each team needing to configure their own connections or negotiate quota independently. An Azure AI Project sits inside the Hub and provides team-level isolation: each project has its own experiments, deployments, prompt flows, evaluations, and fine-tuning jobs, all governed by the Hub ' s shared infrastructure. Different teams get their own project with independent access controls via RBAC, while the platform team manages the shared Hub. This pattern eliminates redundant resource configurations across teams while maintaining clear team-level boundaries — the correct structure for centralized governance with team isolation.

Microsoft Learn Reference Topic: Microsoft Azure AI Foundry hub and project architecture – Centralized governance and team isolation

Step 1: Create an Azure DevOps project

Step 2: Create a release pipeline

Sign in to your Azure DevOps organization and navigate to your project.

Go to Pipelines, and then select New pipeline.

Step 3: Install the Machine Learning extension for Azure Pipelines

You must install and configure the Azure CLI and ML extension.

Step 4: Create a service connection

How to set up your service connection

Select AzureMLWorkspace for the scope level, then fill in the following subsequent parameters.

Note: How to enable model triggering in a release pipeline

Go to your release pipeline and add a new artifact. Click on AzureML Model artifact then select the appropriate AzureML service connection and select from the available models in your workspace.

Enable the deployment trigger on your model artifact as shown here. Every time a new version of that model is registered, a release pipeline will be triggered.

Managed online endpoints with traffic splitting are the best fit because Azure Machine Learning supports multiple deployments behind a single online endpoint and allows production traffic to be distributed between them. This enables a blue-green or progressive rollout strategy : deploy the new model version alongside the current version, validate it independently, send a small percentage of production traffic to it, monitor behavior, and then gradually increase traffic. If problems occur, traffic can be redirected immediately to the previous deployment, providing a practical rollback mechanism.

This approach also aligns with Fabrikam’s requirement to favor managed services and automation because the team has limited Azure DevOps experience. Managed online endpoints handle much of the serving, scaling, security, and monitoring infrastructure automatically and support both traffic splitting and mirrored traffic for safer validation.

Option B can also implement blue-green deployment, but AKS introduces substantially more infrastructure and operational management than Fabrikam requires. Option A would require custom deployment, routing, scaling, and rollback logic. Option D is intended for asynchronous batch inference rather than staged real-time production serving.

Study Guide Reference: Implement machine learning model lifecycle and operations — managed online endpoints, multiple deployments, traffic splitting, blue-green deployment, staged rollout, monitoring, and rollback.

Imagine a hospital analytics firm with data scientists who kick off dozens of training jobs throughout the week. During peak hours, five jobs compete for the same GPU cluster and fail or queue for hours. On quiet nights, that cluster sits completely idle, burning money. Managed compute targets with autoscaling solve both problems: the cluster scales out automatically when multiple jobs arrive simultaneously and scales back to zero when idle. Option A (single shared cluster) is exactly the resource-contention problem Fabrikam already has. Option B (fixed-size cluster) wastes money during off-peak hours. Option C (dedicated per-experiment clusters) eliminates contention but is prohibitively expensive for a cost-conscious healthcare firm. Autoscaling managed compute is the cloud-native answer to variable workload demand.

Microsoft Learn Reference Topic: Azure Machine Learning compute targets – Configure autoscale for compute clusters

The first action should be to evaluate the model output . Fabrikam already has evaluation data containing manually defined expected responses, and the technical requirements explicitly state that the organization must use its existing evaluation datasets based on real data with input-output pairs . Microsoft Foundry supports evaluating a model against an existing curated dataset to establish measurable baseline performance before optimization. Evaluation results can identify weaknesses in accuracy, relevance, groundedness, safety, or other quality dimensions and provide objective evidence for subsequent changes.

This sequencing is critical because Fabrikam also requires that advanced fine-tuning be applied only when prompt engineering is insufficient . Therefore, immediately fine-tuning the model would bypass the required baseline evaluation and optimization process. Fine-tuning should be driven by demonstrated performance gaps, not applied automatically.

Deploying directly to production is also inappropriate because leadership requires low operational risk and gradual rollout. Production traffic should not be used as the first evaluation mechanism when curated evaluation data already exists. Synthetic interaction data is useful when evaluation coverage is insufficient, but Microsoft specifically positions synthetic generation as an option when an adequate test dataset is unavailable. Fabrikam already has suitable evaluation data.

Study Guide Reference: Optimize generative AI systems and model performance — evaluate model performance, benchmark against curated datasets, iterate on prompts, and apply advanced fine-tuning when justified.

Box 1: Regenerate the keys of account1.

Azure Blob Storage support authentication through Account key or SAS token.

To authenticate your access to the underlying storage service, you can provide either your account key, shared access signatures (SAS) tokens, or service principal

Box 2: Update the authentication for store1.

For Azure Machine Learning studio users, several features rely on the ability to read data from a dataset; such as dataset previews, profiles and automated machine learning. For these features to work with storage behind virtual networks, use a workspace managed identity in the studio to allow Azure Machine Learning to access the storage account from outside the virtual network.

Note: Some of the studio ' s features are disabled by default in a virtual network. To re-enable these features, you must enable managed identity for storage accounts you intend to use in the studio.

The following operations are disabled by default in a virtual network:

Preview data in the studio.

Detect when the system is slow at returning output: Latency

Identify when the system cannot sustain expected request volume: Throughput

Measure the duration of handling a request: Response time

Latency measures delay experienced while waiting for model output and is therefore the appropriate metric for identifying degradation in perceived responsiveness. Microsoft describes latency as the time required to obtain a response from the model and exposes metrics such as Time to Response/Time to First Token and Time Between Tokens for generative AI workloads. An increase indicates that users are waiting longer for output to begin or continue.

Throughput represents the amount of workload the system can process over a unit of time. Microsoft describes system-level throughput in terms such as requests per minute and tokens per minute. Consequently, throughput is the correct metric when determining whether a deployment can sustain the required request volume as demand increases.

Response time represents the overall duration required to handle an individual request. It is appropriate for measuring end-to-end request processing and validating request-duration SLOs. Microsoft specifically includes latency, throughput, and response times among the performance metrics candidates are expected to monitor for generative AI applications.

Concurrency measures simultaneous active requests and can influence capacity and latency, but none of these requirements directly asks for simultaneous-request count.

Study Guide Reference: Implement generative AI quality assurance and observability — continuous monitoring, latency, throughput, response times, production troubleshooting, and SLO monitoring.

In Azure Machine Learning, a workspace is the top-level resource that acts as a centralized hub for every ML activity. It stores experiments, pipelines, datasets, models, environments, and compute targets all in one place. Before you can register assets or deploy endpoints, the workspace itself must exist and be shared across the team. Fabrikam ' s core challenge is inconsistent experiment tracking, ad-hoc versioning, and no standardized deployment path — all symptoms of the absence of a single, governed workspace. Option A (registering assets) is a subsequent action only possible after the workspace exists. Option C (online endpoint) is a deployment concern, not an asset-management foundation. Option D (Foundry project) is appropriate for generative AI workloads, not for the traditional ML pipeline standardization described here. The workspace is the prerequisite that makes every other governance action possible.

Microsoft Learn Reference Topic: Azure Machine Learning workspaces – Microsoft Learn: Manage Azure Machine Learning workspaces

In Microsoft Foundry ' s built-in evaluation framework, each evaluator measures a specific dimension of language quality. Fluency evaluates the grammatical correctness, sentence structure, and overall linguistic smoothness of generated text. A high fluency score means the response reads naturally, with proper grammar, punctuation, and sentence construction — regardless of factual accuracy or relevance to the topic. Coherence (option A) measures whether ideas flow logically from one sentence to the next — it is about logical structure, not grammar. Textual Similarity (option B) measures how closely the generated text matches a reference text using metrics like ROUGE or BLEU — it is a comparison metric, not a standalone quality dimension. Groundedness (option C) measures whether claims are supported by the provided context — it is a factual fidelity metric. For grammatical correctness specifically, Fluency is the correct evaluator.

Microsoft Learn Reference Topic: Built-in AI evaluators in Microsoft Foundry – Fluency, Coherence, Groundedness, and Similarity

Fine-tuning is a data-driven process, and you cannot begin configuring or running a fine-tuning job without a properly formatted dataset. Microsoft ' s documentation explicitly states that your training data must be in JSONL format with prompt-completion pairs before you can initiate any fine-tuning job in Microsoft Foundry or Azure OpenAI Studio. Option A (prompt flow templates) is irrelevant at this stage since prompt engineering is a pre-fine-tuning step. Option B (deploying the base model first) is incorrect; fine-tuning trains the weights of the base model without requiring a prior deployment. Option C (enabling tracing in evaluation pipelines) is a post-deployment observability step, not a prerequisite for fine-tuning. Dataset preparation and upload is always the foundational first step in the fine-tuning workflow — without correctly formatted data, no other step can proceed.

Microsoft Learn Reference Topic: Fine-tune models in Azure AI Foundry – Prepare your training dataset

Real-time, predictable: Global Provisioned

Flexible, cost efficient: Global Standard

For the real-time, latency-sensitive workload with predictable global demand , use Global Provisioned . Microsoft Foundry defines Global Provisioned deployments as providing reserved model processing capacity and predictable throughput through provisioned throughput units (PTUs). Because capacity is reserved, this deployment type provides lower and more consistent latency than Global Standard and is specifically recommended for sustained, predictable, high-throughput production workloads.

For the workload requiring flexibility and cost-efficient operation , use Global Standard . Global Standard operates on a pay-per-token model and dynamically routes requests across Azure ' s global infrastructure. Microsoft recommends it as the starting deployment type for most workloads because it offers the lowest price, broad regional availability, and high default quota without requiring the customer to reserve PTU capacity. This makes it appropriate when demand varies and paying for permanently reserved throughput would be inefficient.

The architectural distinction is therefore straightforward: Global Provisioned trades reserved-capacity cost for predictable throughput and reduced latency variance, while Global Standard provides elastic, consumption-based operation with greater latency variability under sustained high load. Microsoft’s deployment overview likewise identifies provisioned throughput for predictable low-latency performance and standard deployment for general hosted workloads.

Study Guide Reference: Design and implement a GenAIOps infrastructure — Microsoft Foundry deployment types, Global Standard, Global Provisioned, PTUs, latency predictability, throughput, and cost optimization.

MLflow is the industry-standard open-source platform for experiment tracking, and Azure Machine Learning has first-class native integration with it. When you use MLflow within an Azure ML job, parameters, metrics, and artifacts are automatically logged to the run history of the AML workspace, making every run reproducible and comparable. Option A (AML job output logs) only captures console output and lacks structured parameter and metric logging. Option C (Application Insights logs) is designed for application-level telemetry, not ML experiment metadata. Option D (Azure Monitor alerts) is a reactive notification tool, not a tracking system. MLflow ' s autologging capability means that for common frameworks such as scikit-learn, XGBoost, and PyTorch, parameters and metrics are captured without a single line of custom code, directly answering Fabrikam ' s requirement for consistent experiment tracking.

Microsoft Learn Reference Topic: Track ML experiments with MLflow – Azure Machine Learning MLflow integration

Domain specialization: Supervised fine-tuning

Poor response accuracy of a RAG-based solution: Apply prompt engineering

For domain specialization , Supervised Fine-Tuning (SFT) is the correct first strategy. Microsoft describes SFT as the foundational fine-tuning technique for training a model from labeled input-output pairs , and specifically identifies domain specialization as one of its principal use cases. Microsoft also recommends starting with SFT for most customization projects because it supports task specialization, instruction following, style, and domain-specific behavior. Fabrikam already possesses evaluation data containing input-output pairs, which aligns directly with the SFT data model.

For poor RAG response accuracy , the first action is prompt engineering . The case explicitly requires advanced fine-tuning only when prompt engineering is insufficient. In a RAG system, prompt engineering determines how the model interprets retrieved context, constrains answers to grounding information, handles missing evidence, and formats responses. Microsoft notes that inadequate RAG prompting can produce false or incomplete answers even when retrieval returns appropriate content.

DPO is primarily appropriate for alignment using preferred versus non-preferred responses, while RFT targets complex reward-based reasoning optimization. Neither is the initial technique for domain specialization in this requirement.

Study Guide Reference: Optimize generative AI systems and model performance — prompt engineering, RAG optimization, supervised fine-tuning, preference optimization, and model customization strategy.