Microsoft AI-300: Operationalizing Machine Learning and Generative AI Solutions

Last checked October 22, 2026 against the official Microsoft exam page.

1. Question 1Implement machine learning model lifecycle and operations

   A data scientist runs many training jobs with different learning rates and needs to compare metrics and parameters side by side. Which tool tracks this?

   1. AA spreadsheet updated by hand after each job
   2. BThe compute instance's desktop wallpaper
   3. CMLflow experiment tracking in the workspace
   4. DDeleting old job outputs to save space
   Show answer and explanationHide answer

   Correct answer: C

   MLflow tracking records parameters, metrics, and artifacts for each run, which allows comparison across jobs. Manual spreadsheets are error-prone, and wallpapers and deletion do not record experiment results.

2. Question 2Implement generative AI quality assurance and observability

   A team evaluates a RAG application and wants to know whether the answers stay within the retrieved sources. Which metric should be used?

   1. ANumber of deployments per month
   2. BGroundedness
   3. CAverage CPU utilization
   4. DDisk input and output rate
   Show answer and explanationHide answer

   Correct answer: B

   Groundedness measures whether the answer's claims are supported by the provided context. Disk rates, CPU, and deployment counts are infrastructure measures and do not assess support.

3. Question 3Optimize generative AI systems and model performance

   A RAG system misses relevant passages because chunks are too large and mix several topics. What is the best first adjustment?

   1. ATune the chunk size and overlap, then measure retrieval relevance on a test set
   2. BIncrease the model's maximum output token length so answers cover missing passages
   3. CRemove the retrieval step and rely on the model's parametric memory
   4. DDouble the number of concurrent users hitting the search index
   Show answer and explanationHide answer

   Correct answer: A

   Chunk size directly affects retrieval precision, and measured relevance shows whether the change helps. Longer outputs, removing retrieval, and more users do not address mixed-topic chunks.

4. Question 4Design and implement an MLOps infrastructure

   A data science team needs its training pipeline to read data from a storage account without any stored keys. Which access configuration fits best?

   1. AA managed identity for the workspace compute with a role assignment on the storage account
   2. BA storage account key hardcoded into the pipeline YAML as a secret value
   3. CA shared access signature valid for ten years embedded in the notebook configuration
   4. DAnonymous public read access enabled on the container so a client can list and download blobs over HTTP
   Show answer and explanationHide answer

   Correct answer: A

   A managed identity with a scoped role assignment gives the compute access without storing secrets. Pasted keys, long-lived SAS tokens, and public containers all expose data or credentials.

5. Question 5Design and implement a GenAIOps infrastructure

   Prompts for a customer assistant are changed often, and the team needs to know which version produced a given result. Which practice provides this?

   1. ASharing one prompt file on a desktop folder copied between several engineers' laptops each week
   2. BStoring prompts in a Git repository with versioned commits and comparing variants
   3. CEditing the prompt text directly in production without a change record
   4. DKeeping prompts only inside a chat thread and session history
   Show answer and explanationHide answer

   Correct answer: B

   Version control gives each prompt a history and lets results be tied to a specific version. Unrecorded production edits, chat threads, and desktop files lose that link.