Find self-service SLURM-as-a-Service options for AI workloads.

Here are the most relevant self-service SLURM-as-a-Service options for AI workloads I found:

• NorthWind Powered SLURM-as-a-Service — Offers fully managed, multi-tenant, on-demand Slurm clusters with self-service access via portal or API, and explicitly supports AI/ML workloads alongside HPC jobs.[1][2]
• Nebius Managed Soperator — A fully managed Slurm-on-Kubernetes solution with self-service availability in Nebius AI Cloud; Nebius also documents self-deployment paths on other Kubernetes platforms and on-premises.[4][15]
• OpenNebula Elastic Slurm Service / OneSlurm Service — Provides a scalable Slurm deployment using preconfigured appliances and OneFlow; it is positioned for AI and HPC workloads on Linux clusters.[5]
• Kobayashi Managed Soperator — Marketed as a fully managed, self-service app for Slurm-based clusters aimed at AI training on NVIDIA GPU clusters.[9]
• StackHPC + Slinky / Azimuth — Not a packaged commercial SaaS in the same way as NorthWind or Nebius, but StackHPC describes self-service portal patterns for isolated Kubernetes clusters and a Slurm cluster for MPI-based training/inference, using Slinky.[3][14]

If you want the closest fit to “self-service SLURM-as-a-Service for AI”, the strongest matches are NorthWind and Nebius Managed Soperator because both explicitly mention self-service, managed delivery, and AI/ML workloads.[1][4][15]

Useful source URLs:

• northwind.co
• nebius.com
• docs.nebius.com
• docs.opennebula.io
• kbytech.com