On Prem Cloud is an R&D initiative to unify multiple OpenStack clusters into a managed, multi-tenant platform.
It delivers Kubernetes-as-a-Service, storage, and AI inference APIs over on-prem infrastructure — optimized for CPU inference to reduce GPU dependency and cost.
Core Features
- Multi-region federation using OpenStack Magnum and Kuryr
- CPU-based inference endpoints for LLM models
- Self-service tenant provisioning with quota isolation
- Ansible-driven automation for reproducible deployments
Tech stack: OpenStack Ansible · Kubernetes · Helm · Ceph · OVN · Terraform · Python