Uploaded September 2026 | Updated November 2026, 3 hours ago
Physical AI systems like robotics and fleets need ML inference pipelines that scale across real-time decisions and offline backfills.
At Ray Summit 2026, Eagle Lo, Hao Yuan, and Freeman Mak from Samsara present Cascade, a declarative inference framework built on Ray that separates pipeline definition from execution. Pipelines are expressed as DSL-defined DAGs, and an execution planner optimizes and schedules them across Ray Serve, Ray Core, and Spark, so the same pipeline runs consistently across online serving and offline batch and backfill workflows, reducing duplicated logic while improving GPU utilization and throughput.
You'll leave with design decisions, tradeoffs, and patterns for building declarative pipelines and unifying online and offline inference.
Liked this video? Check out other Ray Summit breakout session recordings: youtube.com/playlist?list=PLZNaJyYZRPdo
Subscribe to our YouTube channel to stay up-to-date on the future of AI: youtube.com/c/anyscale
๐ Connect with us:
LinkedIn: linkedin.com/company/joinanyscale
X: https://x.com/anyscalecompute
Website: anyscale.com
Physical AI systems like robotics and fleets need ML inference pipelines that scale across real-time decisions and offline backfills.
At Ray Summit 2026, Eagle Lo, Hao Yuan, and Freeman Mak from Samsara present Cascade, a declarative inference framework built on Ray that separates pipeline definition from execution. Pipelines are expressed as DSL-defined DAGs, and an execution planner optimizes and schedules them across Ray Serve, Ray Core, and Spark, so the same pipeline runs consistently across online serving and offline batch and backfill workflows, reducing duplicated logic while improving GPU utilization and throughput.
You'll leave with design decisions, tradeoffs, and patterns for building declarative pipelines and unifying online and offline inference.
Liked this video? Check out other Ray Summit breakout session recordings: youtube.com/playlist?list=PLZNaJyYZRPdo
Subscribe to our YouTube channel to stay up-to-date on the future of AI: youtube.com/c/anyscale
๐ Connect with us:
LinkedIn: linkedin.com/company/joinanyscale
X: https://x.com/anyscalecompute
Website: anyscale.com










