Microsoft Reactor Redmond

Explore the fundamental techniques that power AI applications. This session covers completions and chat flows, function calling to connect AI with external tools, Retrieval-Augmented Generation (RAG) for document-based AI, and responsible AI safety measures. Through hands-on examples you'll learn to implement these patterns in production Java applications.

You'll Learn:

• Streaming completions and conversation management
• Function calling with JSON schemas
• RAG pattern for document Q&A
• Content filtering and safety handling

Observability is about instrumenting AI apps to collect signals that provide insight into model behavior & performance across the GenAIOps lifecycle.

Join us as we talk to Han Che about evaluations, red-teaming & end-to-end observability in Azure AI Foundry – for trustworthy AI!

This hands-on workshop teaches participants to build cost-effective evaluation systems for RAG applications using Azure Cosmos DB's vector search capabilities. Attendees will learn to implement semantic caching techniques that significantly reduce LLM evaluation costs while maintaining fast query performance.

Participants will create a complete evaluation pipeline that measures retrieval quality, answer accuracy, and system performance using industry-standard metrics.

By the end of this session, attendees will have production-ready code and benchmarking tools that can scale across different deployment environments.