Deploying Secure Azure LLM Infrastructure

This unit teaches students how to design, deploy, and secure a three-tier LLM application on Microsoft Azure using managed services. Learners will integrate Retrieval-Augmented Generation workflows and configure Azure Front Door, App Service, API Management, Storage Accounts, and Cosmos DB for scalable, resilient, and secure LLM-driven applications.

Learning Objectives

• Analyze the components, data flows, and security boundaries of a three-tier Azure LLM application, including the OpenAI inference layer, API/service layer, and storage/data layer
• Apply Retrieval-Augmented Generation principles by integrating vector storage and search pipelines with Azure Storage and Cosmos DB to support contextualized model responses
• Demonstrate configuring and securing Azure Front Door, Azure App Service, and API Management to provide resilient, performant, and protected access to LLM APIs (including routing, WAF, authentication, and rate limiting)
• Evaluate Storage Account and Cosmos DB configuration options for performance, consistency, scalability, and compliance, and recommend secure data-handling practices for inference artifacts and embeddings