UCLA Center for Research in Engineering, Media, and Performance and Advanced Research Computing have collaborated to build a set of cloud-based AI inferencing microservices to support an immersive live production of the musical Xanadu. During the production, audience members interact with the production using an app on their mobile phones that allows them to draw images in the air with cell-phone gestures. Those images are then sent to the AI microservices for real-time inferencing. The inferences are then displayed as 3D meshes on stage-based digital panels during the performance. This session will walk participants through the imagining of the performance, the AI microservices we developed, our model selection process, and the cloud architecture. Participants should come away with a good understanding of Immersive Performance with AI, how to create and implement similar AI microservices with AWS and how to build cost consciously.