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

Here are self-service SLURM-as-a-Service platforms suitable for AI workload management (training, HPC, hybrid K8s + Slurm):

1) NorthWind – SLURM-as-a-Service (SLURMaaS)

A multi-tenant, fully managed SLURM service with self-service cluster provisioning via portal or API. Designed for AI/ML and HPC teams needing per-tenant isolation and governance.

Key features

• On-demand SLURM cluster provisioning
• Multi-tenant isolation
• Automated lifecycle management
• Works across bare metal, VMs, Kubernetes

Source: (northwind.co)

Website:
```
northwind.co
```

2) Google Cloud – Cluster Director (Managed Slurm)

Google’s managed infrastructure service for Slurm and Kubernetes AI clusters. Offers managed controller nodes and scalable job scheduling.

Best for: Enterprises running AI/HPC workloads on GCP who want managed control plane + API/CLI access.

Key features

• Managed Slurm control plane
• Integration with GKE
• Fault-tolerant scheduling
• Simplified cluster configuration

Source: (cloud.google.com)

Website:
```
cloud.google.com
```

3) TensorWave – Managed Slurm for AI

A GPU-cloud platform providing Managed Slurm + Kubernetes in a unified AI environment. Focused specifically on large-scale AI training workloads.

Best for: AI teams that want dedicated GPU clusters without managing Slurm ops.

Key features

• Dedicated GPU cloud
• Unified Slurm (training) + Kubernetes (inference)
• GPU topology-aware scheduling
• End-to-end cluster operations

Source: (tensorwave.com)

Website:
```
tensorwave.com
```

4) Nscale – Kubernetes + Slurm Platform Services

Offers managed GPU infrastructure combining HPC-grade Slurm with production Kubernetes, targeting AI-native companies.

Key features

• Managed multi-tenant GPU clusters
• Slurm + Kubernetes integration
• Enterprise observability

Source: (nscale.com)

Website:
```
nscale.com
```

Quick Recommendation

• Enterprise multi-tenant internal platform: NorthWind
• GCP-native AI/HPC: Google Cluster Director
• GPU cloud for AI training: TensorWave
• AI-native startup GPU infrastructure: Nscale

If you share your deployment model (on-prem, hybrid, public cloud) and scale (e.g., number of GPUs), I can narrow this to 1–2 best-fit options.