Memphis Python User Group

This month our speaker is Jandee Richards. Jandee will present Building Production-Ready Agent System with Dapr: From Prototype to Scale.

Making agents production-ready requires solving distributed systems challenges most prototypes ignore. This talk demonstrates how Dapr's distributed systems capabilities including durable workflows, persistent memory, and built-in observability provide the foundation for production-grade multi-agent systems.
We'll compare Dapr's approach to popular agent frameworks like AutoGen, CrewAI, and LangGraph. Through hands-on examples, you'll learn patterns for coordinating agents across services, handling failures gracefully, implementing retry logic, and maintaining audit trails.
The session includes a live demonstration building a multi-agent system showing context persistence, model handoffs, human-in-the-loop approvals, and failure recovery. We'll explore common multi-agent system patterns and how to apply them using Dapr's orchestration capabilities.

Key takeaways: proven orchestration patterns, durability strategies for distributed AI workflows, debugging techniques for agent systems, and a working Dapr agent template you can adapt to your use cases.

Bio:
Jandee Richards is Chief Data Officer at Iridius. She has spent over twenty years building data systems, leading architecture teams, and launching new initiatives across Big Tech, consulting, and industry. She focuses on agentic workflows for intelligent automation in highly regulated industries. She also serves on the University of Washington Analytics advisory board.

PS: The results of the poll came out overwhelmingly in favor of deploying FastAPI to Azure. I'll keep that in the queue for next month if we don't have anyone from the community speak. But I think Jandee will be inspiring.