The Clusterone Deep Learning Platform
Clusterone’s Elastic GPU Scheduler detects unused processors and makes them available to others on the team when working with on-premises boxes. When working in the cloud, Clusterone’s dynamic scheduling and auto-scaling can spin up and spin down resources. This means scientists always have the resources their work requires without incurring billing for idle hardware instances.
The Devopsbay was Clusterone's tech partner and was responsible for tech delivery. The platform frontend, backend, cloud infrastructure and support was done by our team.The primary challenge of the platform was software orchestration using Kubernetes, cloud deployment, and automation. The platform was hosted on-premise and on AWS. Our team also deployed and supported the software for a number of on-prem customers: