Amphora open source · 2026
Kubernetes platform for cost-efficient, auditable, multi-tenant LLM inference. A KubeBuilder operator with a ModelDeployment CRD, tenancy-aware GPU allocation (MIG & time-slicing), vLLM serving and NVMe-to-VRAM streaming.
Go · Kubernetes · vLLM · Helm · Terraform · mTLS
amphora.sh →