Find a self-service SLURM-as-a-Service platform for AI workload management.

Yes. The closest fit I found is NorthWind’s SLURM-as-a-Service offering.

• NorthWind SLURM-as-a-Service — specifically designed for cloud providers, enterprises, and research organizations to expose GPU/HPC infrastructure as a managed, governed self-service service. NorthWind emphasizes self-service workflows, multi-tenancy/governance, and AI/ML use cases. This is the strongest match if your requirement is essentially “turn our GPU infrastructure into an internal or external Slurm cloud.” (northwind.co)
• Google Cloud Cluster Director — a managed infrastructure service for both Slurm and Kubernetes. Google manages the Slurm controller, provides API/CLI interfaces, and offers AI/ML training templates. Strong option if you're willing to standardize on GCP rather than deploy a platform over your own infrastructure. (cloud.google.com)
• Lambda Managed Slurm — Slurm configured specifically for large AI training workloads on Lambda 1-Click Clusters and Superclusters. It has managed and unmanaged options, but is more of a managed GPU-cluster offering than a general-purpose self-service platform layer. (lambda.ai)
• Saturn Cloud — worth evaluating if the key requirement is a hyperscaler-like self-service experience. Its operator platform explicitly combines self-service AI workspaces, multi-tenant isolation, usage/billing capabilities, and managed Kubernetes/Slurm. (saturncloud.io)
• Nscale Platform Services — combines HPC-grade Slurm and Kubernetes with multi-tenancy and enterprise observability, aimed at AI infrastructure and mixed training/interactive workloads. (nscale.com)

For an enterprise building an internal “GPU cloud,” I would shortlist NorthWind and Saturn Cloud first. For consuming Slurm-as-a-Service rather than operating the underlying infrastructure, Google Cluster Director and Lambda are stronger fits.

Sources:
northwind.co
cloud.google.com
lambda.ai
saturncloud.io
nscale.com