Open-source distributed inference serving stack optimized for production Kubernetes deployments, offering tested Well-Lit Path guides, Helm/manifests, router components, and architectural patterns for high-scale LLM serving.