Get Your CertAI-300PlayCompete
‹ AI-300 tests

Microsoft Machine Learning Operations Engineer Associate (AI-300) practice questions

100 questions in 10 tests. Open one to answer it on its own, or take a whole test.

Test 1

Take Test 1 ›
  1. A data scientist currently develops notebooks on an Azure Machine Learning compute instance. The team now needs to run a nightly training job over a much larger dataset. The job… · Design and implement an MLOps infrastructure
  2. Your organization trains models in a development Azure Machine Learning workspace and serves them from a production workspace that sits in a different Azure subscription and… · Design and implement an MLOps infrastructure
  3. A scikit-learn training script runs as an Azure Machine Learning command job. The team wants the job to capture hyperparameters, evaluation metrics, and the trained model… · Implement machine learning model lifecycle and operations
  4. A managed online endpoint serves the blue deployment of a fraud model. A retrained green deployment is ready, and the release engineer wants to observe how it behaves on real… · Implement machine learning model lifecycle and operations
  5. A demand-forecasting model has been live on an Azure Machine Learning managed online endpoint for three months. Ground truth labels arrive months later, so accuracy can't be… · Implement machine learning model lifecycle and operations
  6. A customer-facing copilot built on Microsoft Foundry runs at a steady, well-measured request rate all day. On its standard deployment, response times vary with overall demand and… · Design and implement a GenAIOps infrastructure
  7. An agent application hosted on an Azure App Service calls a Microsoft Foundry project by using an API key stored in the app's configuration. A security review demands three… · Design and implement a GenAIOps infrastructure
  8. Testers reviewing a retrieval-augmented HR assistant report that some answers are well written and clearly on topic, but state leave entitlements that do not appear anywhere in… · Implement generative AI quality assurance and observability
  9. A support agent built in Microsoft Foundry passed its pre-production evaluation against a curated test dataset and is now live. Leadership wants quality and safety scores… · Implement generative AI quality assurance and observability
  10. A support assistant on a Standard deployment sends a 3,500-token prompt on every call. The system message opens with a line carrying the current UTC timestamp and the signed-in… · Optimize generative AI systems and model performance

Test 2

Take Test 2 ›
  1. A compliance review of an Azure Machine Learning workspace requires that the workspace itself hold no storage account keys, SAS tokens, or service principal secrets, and that… · Design and implement an MLOps infrastructure
  2. A regulated team is configuring the managed virtual network of an Azure Machine Learning workspace. Their stated priority is minimizing the risk of data exfiltration, but… · Design and implement an MLOps infrastructure
  3. A sweep job explores 60 configurations of a deep learning model, and each trial logs validation AUC once per epoch. The team wants to cut the compute bill, but a previous sweep… · Implement machine learning model lifecycle and operations
  4. Every night a retailer scores about 40 million customer records that sit as parquet files in a data lake. Nothing waits on an individual prediction (the run only has to finish… · Implement machine learning model lifecycle and operations
  5. A scikit-learn model was logged with mlflow.sklearn.log_model() and registered in the workspace with type mlflow_model. An engineer new to the team has written a score.py and a… · Implement machine learning model lifecycle and operations
  6. A German insurer is deploying a chat model in Microsoft Foundry for an internal claims assistant. Legal requires that prompts and completions be processed only within the EU data… · Design and implement a GenAIOps infrastructure
  7. A Microsoft Foundry project hosts an agent that a line-of-business web application calls at runtime. The web app has its own managed identity. It only ever invokes the agent's… · Design and implement a GenAIOps infrastructure
  8. Before a marketing copy generator ships, the legal team asks for measured evidence that its output does not reproduce copyrighted song lyrics or passages from published books… · Implement generative AI quality assurance and observability
  9. A retrieval-augmented assistant over long policy PDFs indexes them as fixed-size chunks with no overlap. Users report answers that stop halfway through an explanation, and… · Optimize generative AI systems and model performance
  10. A model that has already been supervised fine-tuned on a company's support corpus answers accurately but far too bluntly. Reviewers can reliably say which of two candidate… · Optimize generative AI systems and model performance

Test 3

Take Test 3 ›
  1. Data scientists on a team start every training script with a block of pip install commands. The same script run two weeks apart produced noticeably different evaluation metrics… · Design and implement an MLOps infrastructure
  2. A GitHub Actions workflow deploys an Azure Machine Learning workspace with Bicep and then submits a pipeline job. Security has banned long-lived secrets from the repository, and… · Design and implement an MLOps infrastructure
  3. Three teams each keep their own copy of the same feature-engineering script inside their training pipelines. The copies have drifted apart, and a bug fix one team made never… · Implement machine learning model lifecycle and operations
  4. A business analyst with limited machine learning background needs a baseline classifier over a tabular dataset by the end of the week. They want the service to try several… · Implement machine learning model lifecycle and operations
  5. Clients calling a managed online endpoint start receiving HTTP 429 Too many pending requests during busy periods. The deployment's instances are nowhere near CPU saturation, and… · Implement machine learning model lifecycle and operations
  6. Four product teams at one company will each build their own agents in Microsoft Foundry. Each team's agents, uploaded files, and evaluation results must be kept separate from the… · Design and implement a GenAIOps infrastructure
  7. A regulated production copilot runs on a Standard deployment of a chat model. Compliance requires that the model version behind it never change without a validation run first, so… · Design and implement a GenAIOps infrastructure
  8. A multi-step Foundry agent occasionally returns an answer built from out-of-date information and takes nine seconds instead of the usual two. The team can already see the final… · Implement generative AI quality assurance and observability
  9. A retrieval-augmented assistant over an industrial parts catalog handles conceptual questions well, but when a technician pastes an exact part number such as RX-4410-B, the… · Optimize generative AI systems and model performance
  10. A RAG index built on text-embedding-3-large over roughly 12 million chunks has become the largest cost line in the product. The team must cut vector storage substantially while… · Optimize generative AI systems and model performance

Test 4

Take Test 4 ›
  1. A team trains a churn model on Parquet files that live in two different Azure Storage containers. Before every training run, the same steps must be applied: read both locations… · Design and implement an MLOps infrastructure
  2. An Azure Machine Learning workspace is secured with a private endpoint in the company's virtual network, and its public network access is disabled. Data scientists who work from… · Design and implement an MLOps infrastructure
  3. A team runs a PyTorch DistributedDataParallel training script as an Azure Machine Learning command job on a compute cluster of two nodes, each with 4 GPUs. The job succeeds, but… · Implement machine learning model lifecycle and operations
  4. A workspace has eight registered versions of a fraud model. Versions 1 to 5 are obsolete, and the team wants them to stop appearing when people list models, so nobody deploys one… · Implement machine learning model lifecycle and operations
  5. A churn model in production reads features from a managed feature store. A data scientist wants to add new window aggregates for a second model version and must not change what… · Implement machine learning model lifecycle and operations
  6. A support assistant must reference product codes in its answers. The quality team wants a batch evaluation metric that scores 1.0 only when every product code in a response… · Implement generative AI quality assurance and observability
  7. A team wants to deploy an open-source model from the Hugging Face collection in Microsoft Foundry by using managed compute (preview). The subscription already has plenty of… · Design and implement a GenAIOps infrastructure
  8. An agent's system prompt is edited by several people directly in the portal. Recently a change made the agent's tone worse, and nobody can tell who changed the prompt, when, or… · Design and implement a GenAIOps infrastructure
  9. Four teams share one Microsoft Foundry resource, each working in its own Foundry project, and all use Azure OpenAI models that are sold by Azure. Finance needs each month's token… · Implement generative AI quality assurance and observability
  10. A RAG application uses a pure vector query with k set to 5 against Azure AI Search. When users ask off-topic questions, the search still returns five chunks with low similarity… · Optimize generative AI systems and model performance

Test 5

Take Test 5 ›
  1. Fifteen data scientists must run experiments, submit training jobs to existing compute, and register models in a shared Azure Machine Learning workspace. They must not be able to… · Design and implement an MLOps infrastructure
  2. A security review of an Azure Machine Learning workspace that was created in August 2024 finds that the workspace's system-assigned managed identity holds the Contributor role on… · Design and implement an MLOps infrastructure
  3. A loan-approval classifier has 92 percent overall accuracy, but support staff report that it fails noticeably more often for some kinds of applicants. The team does not know… · Implement machine learning model lifecycle and operations
  4. A team tunes a deep learning model on expensive GPU nodes. The search space includes the learning rate as a continuous log-scale range (loguniform), a dropout rate as a… · Implement machine learning model lifecycle and operations
  5. A model monitor in Azure Machine Learning tracks prediction accuracy against collected ground truth with a threshold (the model performance signal, currently in preview). The… · Implement machine learning model lifecycle and operations
  6. A platform team must deploy a model from the Foundry catalog for a production application. The requirements are: authentication with Microsoft Entra ID instead of API keys, a… · Design and implement a GenAIOps infrastructure
  7. A team manages its Foundry resource with Bicep. Development and production applications must call the same model, and the team wants separate deployments so that quota… · Design and implement a GenAIOps infrastructure
  8. A team has an evaluation dataset whose columns are named user_question, bot_answer and source_docs. The Foundry evaluator they want to run expects the inputs query, response and… · Implement generative AI quality assurance and observability
  9. A RAG application runs a hybrid query (keyword plus vector) with semantic ranking enabled in Azure AI Search. The vector queries use k set to 10, and the team finds that the… · Optimize generative AI systems and model performance
  10. A team fine-tuned five Azure OpenAI candidate models and deployed all of them so that reviewers could compare outputs. Two weeks later, one model was chosen for production, and… · Optimize generative AI systems and model performance

Test 6

Take Test 6 ›
  1. A subscription has a development workspace and a production workspace in the same region. By default both draw on the same subscription quota for a GPU VM family. Large… · Design and implement an MLOps infrastructure
  2. A company created an Azure Machine Learning workspace with a private endpoint in a hub virtual network. Data scientists connect from the corporate network over a VPN, but studio… · Design and implement an MLOps infrastructure
  3. A training job reads a 400 GB image folder registered as a uri_folder data asset. The job YAML sets the input mode to download. Copying the folder to the compute node takes over… · Implement machine learning model lifecycle and operations
  4. A managed online endpoint serves a model from a deployment named blue. The team deployed an improved model as green and wants 10 percent of live requests to be answered by green… · Implement machine learning model lifecycle and operations
  5. A pipeline's first step is a command component that calls a REST API and writes the latest exchange rates to its output. The component's inputs, code and environment never… · Implement machine learning model lifecycle and operations
  6. A retailer's chat assistant is deployed in Microsoft Foundry with the default content filter. The company requires that its responses never contain the names of two specific… · Design and implement a GenAIOps infrastructure
  7. An operations team wants to be notified as soon as a model deployment in Microsoft Foundry starts returning throttling responses, before users complain. Which approach should… · Design and implement a GenAIOps infrastructure
  8. A RAG application often gives poor answers. The team suspects the search step returns irrelevant chunks, and wants a built-in evaluator that isolates the quality of the retrieved… · Implement generative AI quality assurance and observability
  9. Before launching a customer-facing agent, a security team wants an automated and repeatable scan that attacks the target with adversarial prompts by using several attack… · Implement generative AI quality assurance and observability
  10. A chat interface waits for the full model response before showing anything. Long answers take about 12 seconds, and users feel the app has frozen. The answers cannot be… · Optimize generative AI systems and model performance

Test 7

Take Test 7 ›
  1. Security requires that any Azure Machine Learning workspace created in a subscription without public network access disabled is automatically changed to disable it, so that no… · Design and implement an MLOps infrastructure
  2. A team submits occasional training jobs and wants to avoid creating, sizing and managing compute clusters. They want the lowest administrative effort while jobs still run on… · Design and implement an MLOps infrastructure
  3. A managed online endpoint runs a deployment with a fixed instance count of 2. CPU usage spikes every weekday morning and latency rises. The team wants more instances to be added… · Implement machine learning model lifecycle and operations
  4. A large model needs about 10 minutes to load inside its container. When deployed to a managed online endpoint with the default probe settings, the deployment fails and the logs… · Implement machine learning model lifecycle and operations
  5. A team wants its training pipeline job to run every Monday at 6 AM without anyone starting it manually. Which approach should they use? · Implement machine learning model lifecycle and operations
  6. Several internal apps call one model deployment in Microsoft Foundry through Azure API Management. One app sometimes consumes most of the deployment's tokens, and the other apps… · Design and implement a GenAIOps infrastructure
  7. A regulated firm requires that Foundry Agent Service store conversation threads, uploaded files and vector search data in resources that the firm owns in its own subscription… · Design and implement a GenAIOps infrastructure
  8. A customer support agent sometimes misunderstands what the user is asking for, and then calls tools for the wrong task. The team wants a built-in agent evaluator that scores… · Implement generative AI quality assurance and observability
  9. A RAG assistant summarizes emails and web pages. An attacker hides text in a retrieved document that tells the model to ignore its system prompt and reveal confidential data… · Implement generative AI quality assurance and observability
  10. A company classifies 2 million support tickets every night. Results are needed within 24 hours, and the team wants the lowest cost without consuming the quota used by its… · Optimize generative AI systems and model performance

Test 8

Take Test 8 ›
  1. A team creates a GPU compute cluster with a maximum of 4 nodes of a specific VM family. Creation fails with an error that there is not enough quota for that VM family in the… · Design and implement an MLOps infrastructure
  2. A company must train models on GPUs in its own on-premises Kubernetes cluster because of data residency rules, yet wants to submit and track jobs from its Azure Machine Learning… · Design and implement an MLOps infrastructure
  3. A loan approval model is accurate overall, but the team sees that approval rates differ noticeably between groups defined by a sensitive attribute. In the Responsible AI… · Implement machine learning model lifecycle and operations
  4. The team wants to set up model monitoring for an online deployment, but the deployment does not store any production data. They need the inputs and outputs of scoring requests to… · Implement machine learning model lifecycle and operations
  5. A custom scoring script for a managed online endpoint loads the model file from disk inside its run() function. Every request now takes several seconds. Where should the model be… · Implement machine learning model lifecycle and operations
  6. An application uses model deployments in two Azure regions through Azure API Management. When the primary deployment returns 429 or 5xx responses, the team wants requests to fail… · Design and implement a GenAIOps infrastructure
  7. A chatbot and a nightly summarization job share one model deployment. When the summarizer runs, the chatbot receives 429 errors. The team must guarantee that the summarizer can… · Design and implement a GenAIOps infrastructure
  8. A security team needs a record of who created, changed or deleted model deployments and access settings on a Foundry resource, so that it can be queried in Log Analytics. Which… · Implement generative AI quality assurance and observability
  9. A team has an evaluation dataset with a query, the model's response and the retrieved context for each row, but no reference answers. Which set of built-in evaluators can they… · Implement generative AI quality assurance and observability
  10. A company's product documentation changes every week. Its assistant must answer with current information and point to the source documents. The team is considering fine-tuning a… · Optimize generative AI systems and model performance

Test 9

Take Test 9 ›
  1. A compute cluster has a minimum of 0 nodes and an idle time before scale down of 120 seconds. The team submits training jobs about every 5 minutes during the day, and every job… · Design and implement an MLOps infrastructure
  2. A compliance team requires that operations on an Azure Machine Learning workspace, such as job submissions and compute changes, are logged and retained centrally for one year… · Design and implement an MLOps infrastructure
  3. A retailer wants to predict daily demand for the next 14 days, separately for each of 300 stores, by using automated ML. The data has a date column and a store identifier column… · Implement machine learning model lifecycle and operations
  4. Every time the team fixes the scoring script of an online deployment, they wait many minutes for a cloud deployment just to see the next error. Which practice gives a faster… · Implement machine learning model lifecycle and operations
  5. A developer trains models on a laptop with a script that uses MLflow. The runs are saved only on the laptop, but the team wants the metrics and parameters to appear in the Azure… · Implement machine learning model lifecycle and operations
  6. A Foundry application uses its managed identity to query an Azure AI Search index. API key access on the search service is disabled, and queries return 403 Forbidden. The team… · Design and implement a GenAIOps infrastructure
  7. A team fine-tuned five candidate models and wants reviewers to test each one for a few hours to pick a winner. The candidates will not serve production traffic, and the team… · Design and implement a GenAIOps infrastructure
  8. A team runs the Violence risk and safety evaluator, with its default settings, on three responses that receive severity scores of 1, 3 and 5. Which results are reported as fail? · Implement generative AI quality assurance and observability
  9. A team changes the prompts of its assistant several times a week. They want to stop a change that lowers answer quality from being merged. Which approach should they adopt? · Implement generative AI quality assurance and observability
  10. A team uses a reasoning model to sort support tickets into five simple categories. Latency and token cost are high, because the model produces long internal reasoning for a… · Optimize generative AI systems and model performance

Test 10

Take Test 10 ›
  1. A workspace's Azure Container Registry is behind a private endpoint, and environment image builds fail because the default build compute cannot reach the registry. The security… · Design and implement an MLOps infrastructure
  2. Developers often leave compute instances running overnight, sometimes with a notebook kernel still open. The team wants every compute instance to stop at 8 PM on weekdays, even… · Design and implement an MLOps infrastructure
  3. Client applications call a managed online endpoint by using managed identities. Security forbids shared keys, and wants the scoring permission to be checked by Azure RBAC and… · Implement machine learning model lifecycle and operations
  4. An upstream system change causes some input features to arrive as null, with the wrong data type, or outside the values seen in training. The team wants the model monitor to… · Implement machine learning model lifecycle and operations
  5. A sweep job is configured for 40 trials on a cluster that the team wants to limit to 4 trials running at a time, to stay within quota. Each trial runs on one node. Which sweep… · Implement machine learning model lifecycle and operations
  6. Company policy requires the data of a Foundry resource to be encrypted with a customer-managed key. Which requirement applies to the key store that holds the key? · Design and implement a GenAIOps infrastructure
  7. A team deployed the gpt-4.1 model under the deployment name support-chat. Their application code passes model="gpt-4.1" in each request and receives a 404 DeploymentNotFound… · Design and implement a GenAIOps infrastructure
  8. A team hosts agents in Foundry Agent Service and wants to see traces of their runs, including tool calls and model calls, with the least effort and no application code changes… · Implement generative AI quality assurance and observability
  9. An Application Insights resource is connected to a Foundry project and traces are being collected, but an engineer who has access to the project cannot query the telemetry in the… · Implement generative AI quality assurance and observability
  10. During fine-tuning, the training loss keeps falling, but the validation loss starts to rise after epoch 3 of 6. The model's answers on new prompts are getting worse. Which change… · Optimize generative AI systems and model performance