Multi-Model Composition with Ray Serve Deployment Graphs

Explore multi-model composition using Ray Serve deployment graphs in this 30-minute talk from Anyscale. Discover how to leverage Ray's programmable and general-purpose distributed computing capabilities to author, orchestrate, scale, and deploy complex serving graphs as a DAG under a unified set of APIs. Learn to program multiple models dynamically on your laptop as if writing a local Python script, deploy to production at scale, and perform individual upgrades. Gain insights into streamlining microservice-like functionality for efficient model deployment and management.