Open-source Kubernetes-native stack for distributed LLM inference. The project runs engines such as vLLM and SGLang across a cluster and turns single-node engines into production-grade serving on existing hardware.