AI-300 study guide · 2026

The 5 AI-300 domains, explained

5 domains and 14 official functional groups. Implement machine learning model lifecycle and operations carries the most weight at 25–30%. Each guide explains what the domain covers, how its questions are framed and where candidates lose marks.

Start with Domain 2 See all domains
Questions
Not published
Time
120 minutes
Passing score
700 / 1000
Outline
AI-300 study guide on Microsoft Learn, page updated October 15, 2026 (current)

Exam blueprint

Bar widths follow the official weights, scaled to fit the bar. Weight ranges are drawn at their midpoint.

  1. 1 15–20%
  2. 2 25–30%
  3. 3 20–25%
  4. 4 10–15%
  5. 5 10–15%
  1. 1 Design and implement an MLOps infrastructure
  2. 2 Implement machine learning model lifecycle and operations
  3. 3 Design and implement a GenAIOps infrastructure
  4. 4 Implement generative AI quality assurance and observability
  5. 5 Optimize generative AI systems and model performance

All 5 domains

14 official functional groups 60–78 hrs suggested study

  1. Domain 01 15–20%

    Design and implement an MLOps infrastructure

    Building the Azure Machine Learning foundation for MLOps: workspaces, datastores, compute targets and access control, versioned data assets, environments and components, registries, and infrastructure as code with Bicep, Azure CLI, GitHub Actions and Git.

    Inside this domain

    • Create and manage resources in a Machine Learning workspace
    • Create and manage assets in a Machine Learning workspace
    • Implement IaC for Machine Learning

    Watch for Compute instance for production retraining

    Functional groups
    3
    Questions est.
    ~9
    Study time est.
    12–16 hrs
  2. Domain 02 25–30%

    Implement machine learning model lifecycle and operations

    The largest AI-300 skill area: tracking and tuning training with MLflow, AutoML and sweep jobs, registering and evaluating models responsibly, deploying to online or batch endpoints with safe rollout, and monitoring drift to trigger retraining.

    Inside this domain

    • Orchestrate model training
    • Implement model registration and versioning
    • Deploy machine learning models for production environments
    • Monitor and maintain machine learning models in production

    Watch for Mirroring as rollout

    Functional groups
    4
    Questions est.
    ~14
    Study time est.
    18–24 hrs
  3. Domain 03 20–25%

    Design and implement a GenAIOps infrastructure

    Setting up Microsoft Foundry for production generative AI: resources and projects, managed identities and RBAC, private networking and Bicep, choosing and deploying foundation models, provisioned throughput, and managing prompts and variants under Git version control.

    Inside this domain

    • Implement Foundry environments and platform configuration
    • Deploy and manage foundation models for production workloads
    • Implement prompt versioning and management with source control

    Watch for Global for everything

    Functional groups
    3
    Questions est.
    ~11
    Study time est.
    14–18 hrs
  4. Domain 04 10–15%

    Implement generative AI quality assurance and observability

    Proving generative AI applications and agents work and keep working: building test datasets and column mappings, quality and safety evaluators, automated evaluation in pipelines, and monitoring latency, tokens, cost and traces in Microsoft Foundry.

    Inside this domain

    • Configure evaluation and validation for generative AI applications and agents
    • Implement observability for generative AI applications and agents

    Watch for Relevance catches hallucination

    Functional groups
    2
    Questions est.
    ~6
    Study time est.
    8–10 hrs
  5. Domain 05 10–15%

    Optimize generative AI systems and model performance

    Improving generative AI quality and efficiency after launch: tuning chunking, thresholds and retrieval strategies, choosing embedding models, applying hybrid search, measuring RAG with relevance metrics and A/B tests, and running fine-tuning from synthetic data to production.

    Inside this domain

    • Optimize retrieval-augmented generation (RAG) performance and accuracy
    • Implement advanced fine-tuning and model customization

    Watch for Fine-tune for fresh facts

    Functional groups
    2
    Questions est.
    ~6
    Study time est.
    8–10 hrs

Study the heaviest domains first

Marks follow weight. This order puts the most marks behind you first, across about 60–78 hrs of study.

  1. 2 Implement machine learning model lifecycle and operations 25–30% · 18–24 hrs
  2. 3 Design and implement a GenAIOps infrastructure 20–25% · 14–18 hrs
  3. 1 Design and implement an MLOps infrastructure 15–20% · 12–16 hrs
  4. 4 Implement generative AI quality assurance and observability 10–15% · 8–10 hrs
  5. 5 Optimize generative AI systems and model performance 10–15% · 8–10 hrs

Try one

Practice questionDomain 2: Implement machine learning model lifecycle and operations

A managed online endpoint serves model version 4 through a deployment named blue. You created a green deployment with version 5 and want to compare its predictions on real requests for a week without any customer receiving a version 5 response. What should you configure?

One trap from every domain

The answer that looks right, and the clue that gives it away.

  • Domain 1

    Compute instance for production retraining

    Clue Words such as nightly retraining, several jobs or scale out.

  • Domain 2

    Mirroring as rollout

    Clue The question asks for users to be served by the new version.

  • Domain 3

    Global for everything

    Clue Mentions of residency, sovereignty or processing location.

  • Domain 4

    Relevance catches hallucination

    Clue Answers mention facts missing from the source documents.

  • Domain 5

    Fine-tune for fresh facts

    Clue Frequently updated documents or a need for citations.

Other Artificial Intelligence & Machine Learning certifications

Study guides for exams that overlap with AI-300.

Practise what you read

AI-300 domains: common questions

How many domains are in the AI-300 exam?

AI-300 has 5 domains with 14 official functional groups between them: Design and implement an MLOps infrastructure, Implement machine learning model lifecycle and operations, Design and implement a GenAIOps infrastructure, Implement generative AI quality assurance and observability and Optimize generative AI systems and model performance.

Which AI-300 domain carries the most weight?

Implement machine learning model lifecycle and operations is the heaviest at 25–30%, followed by Design and implement a GenAIOps infrastructure (20–25%). The lightest is Optimize generative AI systems and model performance (10–15%). Weights come from Microsoft's official outline.

What is the AI-300 exam format?

Questions: Not published. Time allowed: 120 minutes. Passing score: 700 / 1000.

How long should I study for AI-300?

Our planning estimate is 60–78 hrs across all domains, split by weight. It is an editorial estimate, not an official figure, so adjust it to your experience and to how you score on practice questions.

Where should I start?

Start with Implement machine learning model lifecycle and operations, because it carries the most marks, then work down by weight. Use the checklist on each domain page to confirm you can do what the domain asks before moving on.

Which outline is this guide based on?

This guide follows AI-300 study guide on Microsoft Learn, page updated October 15, 2026 (current), published by Microsoft. It was last reviewed in October 2026.

Domain names, weights and functional groups come from Microsoft's official outline (AI-300 study guide on Microsoft Learn, page updated October 15, 2026 (current)). Question counts and study hours are editorial planning estimates, not official figures. Last reviewed October 2026.