Plan and Manage an Azure AI Solution¶

Overview¶

This domain covers selecting appropriate AI services, planning security configurations, and creating and managing AI service resources. It represents 15-20% of the exam and focuses on foundational decisions for AI solution architecture.

Key Topics¶

Select Appropriate AI Service¶

Computer vision service selection
Natural language processing service selection
Speech service selection
Generative AI service selection
Understanding service capabilities and limitations
Choosing between custom and pre-built models

Plan and Configure Security¶

AI service security requirements planning
AI service security configuration
Access key and endpoint management
Diagnostic logging configuration
Network security for AI services
Data privacy and compliance considerations

Create and Configure AI Service¶

AI service resource creation
Endpoint configuration
Keys and endpoint URL management
Cost management and optimization
AI service monitoring and metrics
Resource scaling and performance tuning

Services Reference¶

Core AI Services¶

Azure AI Vision (Computer Vision)
Azure AI Language
Azure AI Speech
Azure OpenAI Service
Azure AI Document Intelligence
Azure Cognitive Search

Supporting Services¶

Azure Key Vault (for key management)
Azure Monitor (for monitoring)
Log Analytics
Azure Virtual Networks
Azure Private Link
Managed Identity

Best Practices¶

Service Selection¶

Use pre-built models when possible for faster deployment
Choose custom models when domain-specific accuracy is needed
Consider latency requirements for real-time scenarios
Evaluate cost vs. performance trade-offs
Plan for scalability and growth
Consider regional availability of services

Security Implementation¶

Store API keys in Azure Key Vault
Use managed identities instead of keys when possible
Implement network restrictions with Private Link
Enable diagnostic logging for audit trails
Apply least privilege access principles
Rotate keys regularly

Resource Management¶

Use appropriate pricing tiers for workload
Implement cost alerts and budgets
Monitor usage patterns and optimize
Use single-region vs multi-region strategically
Tag resources for cost tracking
Plan for disaster recovery

Performance Optimization¶

Choose regions close to users for low latency
Use appropriate batch sizes for throughput
Implement caching for repeated requests
Monitor throttling and adjust capacity
Use async operations for better scalability
Optimize request payloads

Common Scenarios¶

Service Deployment¶

Multi-region deployment for high availability
Development, test, and production environment setup
Shared services across multiple applications
Isolated services for compliance requirements
Hybrid on-premises and cloud deployments

Security Configuration¶

Private endpoint setup for enhanced security
Managed identity for secure service-to-service calls
Key rotation without application downtime
Audit logging for compliance requirements
Data residency compliance for regulated industries

Cost Optimization¶

Choosing right pricing tier based on usage
Implementing caching to reduce API calls
Batching requests to improve efficiency
Using commitment tiers for predictable workloads
Monitoring and alerting on cost anomalies

Monitoring and Troubleshooting¶

Setting up Application Insights integration
Creating custom metrics for business KPIs
Configuring alerts for service health
Analyzing usage patterns and trends
Troubleshooting API errors and latency issues

Study Tips¶

Understand capabilities and limitations of each AI service
Learn when to use custom vs pre-built models
Practice creating AI services via Portal, CLI, and SDKs
Study security best practices for AI services
Understand pricing models and cost optimization strategies
Learn about managed identities and their benefits
Practice configuring diagnostic settings
Study Private Link and VNet integration
Understand regional availability and data residency
Learn monitoring and alerting configuration