Combining the ease of use of Ray with the power of YellowDog’s fully-featured dynamic workload manager to deploy Ray clusters anywhere, at scale, and to budget.
London, UK – October 14, 2025 — YellowDog, the industry leader in large-scale compute infrastructure, today announced its release of RayDog, YellowDog’s open-source integration to Ray.io. With RayDog, Ray users get seamless access to YellowDog’s unmatched ability to build compute clusters for distributing workloads at massive scale.
“Our mission has always been to give organisations easy access to compute at scale, helping them accelerate results and make faster decisions” said Bruce Beckloff, CEO at YellowDog. “With our new RayDog integration, we’re extending that advantage to every Ray user — combining the simplicity and ease of use of Ray with YellowDog’s exceptional and proven ability to scale across massive clusters, on premises or in the cloud.”
Ray: the AI/ML compute engine
Ray is an open-source framework that accelerates AI and machine learning (ML) development by building on familiar Python and ML workflows. Developers can prototype models and applications locally, then scale them seamlessly to large clusters in the cloud. By removing infrastructure friction, Ray enables teams to iterate faster, train AI models efficiently, and deliver machine learning solutions at scale with agility.
RayDog: Scaling Ray to new heights
RayDog takes Ray to the next level by combining its ease of use with YellowDog’s powerful orchestration platform. Engineering teams can rapidly spin up massive clusters across on-premises and cloud resources, multiple regions, and even different cloud providers — ensuring the right compute is always available. Teams struggling with the complexities of scaling infrastructure can turn to YellowDog for a simpler path. With RayDog, clusters scale to thousands of nodes in minutes and release just as quickly to control costs, while resilient use of Spot instances ensures maximum performance within budget.
“In quantitative finance, time to market, speed and scale are everything,” said Jon Fautley, Head of Cloud Infrastructure at Qube Research & Technologies (QRT). “Ray has become a popular choice for its ease of use and excellent developer experience. Using RayDog to combine it with YellowDog’s powerful compute infrastructure, we can write and test new data processing applications on local datasets, then seamlessly scale them to massive compute power with no code changes. This gives us the power to instantly turn up the dial up when performance matters most — ensuring we hit our objectives while giving our engineers the freedom to innovate faster.”
RayDog is publicly available via PyPI and GitHub, and technical documentation is provided via the YellewDog developer portal.
Try RayDog Today
Get started quickly with RayDog and experience the power of YellowDog’s high-scale compute platform. Our special Getting Started package includes a RayDog license with access to the YellowDog Platform and $500 in AWS credits to take your workloads for a test drive. Learn more on how RayDog combines the ease of Ray with the power YellowDog. Learn more here.
About YellowDog
YellowDog empowers organisations to unlock the full potential of cloud computing by providing high-scale, dynamic workload management solutions that maximise performance while minimising costs. Leading financial institutions, AI innovators, and research organisations rely on YellowDog’s high-performance computing platform to power next-generation workloads. By combining scale, speed, and cost efficiency, YellowDog transforms the economics and efficiency of running large, computationally intensive workloads, giving businesses the ability to make faster, smarter decisions.
Contact and further information:
Richard Collins, Head of Marketing, YellowDog
Tel: +44 7958559579
E: richard.collins@yellowdog.ai
