Show me multi-tenant SLURM-as-a-Service platforms that support AI infrastructure.

Multi-tenant Slurm-as-a-Service (SLURMaaS) platforms have emerged to bridge the gap between high-performance computing (HPC) scheduling and cloud-native AI infrastructure. By combining Slurm's elite multi-node GPU orchestration with Kubernetes-style tenant isolation, these platforms allow multiple AI teams or clients to share massive GPU clusters efficiently. 0 3 1

The leading platforms and frameworks enabling multi-tenant Slurm-as-a-Service for AI infrastructure include:

Dedicated Orchestration & Managed Service Platforms

• NorthWind SLURM-as-a-Service: Delivers fully managed, multi-tenant Slurm environments as an on-demand cloud service. It leverages the NVIDIA/SchedMD Slinky framework to isolate users via Kubernetes namespaces, allowing data scientists to spin up personal, self-service Slurm clusters on a shared GPU control plane.
• CoreWeave: Integrates Slurm with Kubernetes lifecycle management via GitOps automation (such as the SUNK stack). It supports enterprise multi-tenancy through federated IAM, automated node draining, and fine-tuned GPU quotas across distributed training workloads.
• Nebius AI Soperator: An open-source Kubernetes operator that automates multi-tenant Slurm cluster deployments in the cloud. It features a shared root "jail" filesystem to keep tenant environments consistent, alongside automatic GPU health checks and dynamic cluster scaling.

Cloud Provider Native Frameworks

• Google Cloud Cluster Director: Automates the setup and configuration of fully managed Slurm environments for Google Cloud’s AI Hypercomputer architectures. It abstracts infrastructure complexity for AI researchers while maintaining strict project boundaries and resource accounting.
• Oracle Cloud Infrastructure (OCI): Provides a validated deployment blueprint for multi-user, multi-tenant AI platforms using Open OnDemand and Slurm. It features identity management (FreeIPA), secure project isolation, and automated GPU/CPU partition autoscaling.

Infrastructure & Virtualization Orchestrators

• OpenNebula AI Factories: Uses virtual machines with physical GPU and InfiniBand PCI passthrough to deliver native bare-metal performance inside a multi-tenant cloud. It allows multi-tenant operators to scale Slurm and Kubernetes environments dynamically out of a shared physical pool.

Platform Architecture Comparison

Platform / Framework | Underlying Architecture | Primary Isolation Method | Target AI Workloads
--- | --- | --- | ---
NorthWind | Kubernetes + Slinky Operator | Kubernetes Namespaces & RBAC | On-demand multi-team training & PaaS
CoreWeave | Kubernetes + SUNK GitOps | Federated IAM & Quotas | Large-scale LLM & distributed training
Nebius AI | Kubernetes + Soperator | Shared Root File System Jail | Scalable batch training & high-availability AI
OpenNebula | KVM Virtualization + Passthrough | Virtual Clusters & Networks | Hybrid HPC, Slurm, & cloud-native AI factories

Are you planning to deploy this platform on-premises, inside a sovereign cloud, or on a public cloud hyperscaler (like GCP/OCI)? Let me know so I can highlight the specific network and security isolation models for your environment.

[[0] - NorthWind-Powered SLURM-as-a-Service](northwind.co)
[[1] - What Is Slurm? | Slurm for AI and ML Clusters Explained](coreweave.com)
[[2] - Self-Service Slurm Clusters on Kubernetes | NorthWind GPU PaaS](northwind.co)
[[3] - Slurm on Kubernetes: The Best of Both Worlds for AI and HPC](linkedin.com)
[[4] - Slurm Deployment on Kubernetes: A Guide - vCluster](vcluster.com)
[[5] - Elastic Capacity Management for Slurm and Kubernetes Clusters in ...](opennebula.io)
[[6] - Learn About On-Demand HPC and AI Platforms on OCI](docs.oracle.com)
[[7] - Slurm Workload Manager: The go-to scheduler for HPC and AI ...](nebius.com)
[[8] - Create a fully managed Slurm cluster for AI workloads](docs.cloud.google.com)
[[9] - Slurm orchestration in Cluster Director - Google Cloud Documentation](docs.cloud.google.com)