AI-300

Machine Learning Operations Engineer Associate

Microsoft Certified: Machine Learning Operations Engineer Associate

PS academy:/courses> Get-Certification -Id AI-300 | Select-Object -ExpandProperty Overview

40–60

Questions

~120 min

Duration

700

Passing score / 1000

5

Domains

Associate

Level

PSacademy:/courses/ai-300> Get-DomainWeights | Format-Chart

MLOps Infrastructure Design and Implementation

15–20% of exam

ML Model Lifecycle and Operations

25–30% of exam

GenAIOps Infrastructure

20–25% of exam

GenAI Quality Assurance and Observability

10–15% of exam

GenAI Systems and Performance Optimization

10–15% of exam

PSacademy:/courses/ai-300> Get-LearningModules -Cert AI-300

MOD 01

MLOps Infrastructure Design

Azure ML workspace provisioning with Bicep and Terraform IaC
RBAC role assignments: AzureML Data Scientist, Compute Operator
Managed VNet, private endpoints, and outbound rules for compute
Compute clusters, instances, and serverless compute configuration
Data assets, datastores, and feature stores registration
Environment management: Docker images, Conda specs, curated envs

MOD 02

ML Model Lifecycle and Operations

Pipeline components, parallel run steps, and sweep jobs
MLflow experiment tracking: autolog, metrics, artifacts, and tags
Model registry: versioning, stages (staging/production), and lineage
Online endpoints: blue/green traffic split and canary rollout
Batch endpoints for large-scale asynchronous scoring
Data drift detection and model performance degradation monitoring

MOD 03

GenAIOps Infrastructure

Azure AI Foundry Hub and Project provisioning via IaC
PTU (Provisioned Throughput Units) vs. serverless token deployment
Prompt versioning, source control, and flow deployment
API Management (APIM) gateway for LLM traffic routing and throttling
Quota management, capacity planning, and PTU reservation
CI/CD pipelines for prompt flow and model deployment automation

MOD 04

GenAI Quality Assurance and Observability

AI evaluation metrics: groundedness, relevance, coherence, fluency, safety
Azure AI Evaluation SDK and custom evaluator implementation
CI/CD quality gates with automated evaluation thresholds
Distributed tracing with OpenTelemetry and Application Insights
Token usage, latency, and error rate monitoring dashboards
Harmful content risk scoring and responsible AI compliance gates

MOD 05

GenAI Systems and Performance Optimization

RAG pipeline tuning: chunk size, overlap, embedding model selection
Hybrid and vector search optimization in Azure AI Search
Supervised fine-tuning and DPO for domain adaptation
Batch inference and async processing for high-throughput workloads
Semantic caching to reduce redundant LLM calls and cost
Cost optimization: model tiering, caching strategies, and token budgets

Ready to test your AI-300 knowledge?

Practice all domains, run an exam simulation, or drill into your weak areas.