Scaling Massive Transformer Training with KubeRay | Capital One | Ray Summit 2026 @anyscale
Scaling Massive Transformer Training with KubeRay | Capital One | Ray Summit 2026  @anyscale
Uploaded September 2026 | Updated November 2026, 3 days ago
When infrastructure is siloed, I/O overhead can stretch development cycles into days. Capital One processed 3.5TB of data and cut hyperparameter optimization from days to hours, a 16x acceleration in model tuning.

At Ray Summit 2026, Mauilin Patel, Raja Chawat, and Yifei Wang from Capital One showcase how KubeRay on their compute platform bridges the gap between raw data and optimized models with a unified ecosystem of Ray Data, Ray Train, and Ray Tune, enabling users to distribute PyTorch training while eliminating costly I/O bottlenecks. They share platform-level lessons on Kubernetes autoscaling, resource isolation, and memory optimization for large-scale transformer workloads, grounded in real Capital One use cases.

You'll leave with patterns for consolidating fragmented training stacks.

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

đź”— Connect with us:
LinkedIn: linkedin.com/company/joinanyscale
X: https://x.com/anyscalecompute
Website: anyscale.com
Scaling Massive Transformer Training with KubeRay | Capital One | Ray Summit 2026How Workday Achieved 50x Cheaper Model Serving with Ray Serve | Ray Summit 2025From Reservation to 1,600 GPUs: A Petabyte-Scale Ray Data Pipeline | CoreWeave | Ray Summit 2026How Prime Intellect Builds Scalable Infrastructure for Agentic RL | Ray Summit 2025How KubeRay Is Evolving for Massive AI Workloads  | Ray Summit 2025Production-Ready Distributed Inference with Ray Serve | Anyscale | Ray Summit 2026Petabyte-Scale Video Curation on CoreWeave | CoreWeave | Ray Summit 2026Ray Direct Transport: RDMA Support in Ray Core | Ray Summit 2025Building LLaMA: Metas Director of GenAI Sergey Edunov | Ray Summit 2024Maximizing Compute Efficiency on Anyscale | Ray Summit 2025The PARK Stack: The LAMP Stack of the AI Era | Ben LoricaProduction-ready Distributed Inference with Ray Serve
Anyscale |

Scaling Massive Transformer Training with KubeRay | Capital One | Ray Summit 2026

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER