Key Features
Controller + login node pre-configured; GPU compute nodes enrolled via Ansible. Standard SLURM 23.x CLI out of the box.
Queues for H100, B200, and A6000 nodes; fair-share scheduling enabled. No backfill or pre-emption at MVP.
Submit a request, and we add or remove nodes. Hours, not weeks. Pay only for reserved GPUs.
NFS home/project space plus local NVMe scratch. Parallel file system and object storage are roadmap items.
Prometheus + Grafana dashboards; Buzz ops receive hardware alerts and swap failing nodes automatically.
VPN-isolated cluster; Unix user/group separation. Optional identity integration coming soon.
HPC veterans on call (9 × 5) with 24 × 7 hardware escalation.
Use Cases
Large Model Inference
Run massive models with predictable latency. Optimize for throughput, batch size, and performance per watt.
Generative AI applications for text, image, and audio.
Scaling ML infrastructure as your customer base grows.
Bare-metal GPU horsepower, zero scheduler upkeep, and people who speak SLURM fluently. It’s the shortest path from research idea to results—no data-center build-out required.
