# ray-project/kuberay

A toolkit to run Ray applications on Kubernetes

Repository: https://github.com/ray-project/kuberay
Canonical: https://ross.abutalabs.com/products/kuberay
Language: Go
License: Apache-2.0
License Family: permissive
Topics: machine-learning, kubernetes, apache, ray, deep-learning
Last push: 2026-08-24T21:31:28+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 86, longevity 100
- inputs: {"age_days": 2134, "days_push": 9, "days_rel": 13, "gap_med": 35, "n_releases_24m": 14}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2650, forks 830 (observed 2026-08-28T04:07:06.972537+00:00)

## What it is
KubeRay is a Kubernetes operator and toolkit for deploying and managing Ray applications on Kubernetes. It provides RayCluster, RayJob, and RayService custom resources that handle cluster lifecycle, autoscaling, job submission, and zero-downtime serving upgrades.

## Use cases
- run ray clusters on kubernetes
- autoscale distributed machine learning training jobs
- submit ray jobs to kubernetes and clean up after completion
- serve ray serve models with zero-downtime upgrades
- manage fault-tolerant ray workloads on k8s
- deploy llm inference with ray on kubernetes

## When to choose
- you already run Ray workloads and want them managed on Kubernetes
- you need autoscaling, fault tolerance, or zero-downtime upgrades for Ray clusters
- you want declarative CRDs for Ray cluster, job, and serving lifecycle

## When to avoid
- you don't use Kubernetes or Ray
- you need a simple single-node Ray setup without orchestration
- you want a fully managed Ray service rather than self-managed operator

## Facets
- artifact type: infra-config
- maturity: active
- function: container-orchestration, deployment, machine-learning, deep-learning, llm-inference, scheduling, workflow-automation
- domain: machine-learning, deep-learning, cloud-computing, infrastructure-as-code, microservices
- platform: cloud, go, self-hosted
- tags: ray, kubernetes-operator, crd, raycluster, rayjob, rayservice, autoscaling, distributed-computing, kubectl-plugin, containers, devops, kubernetes, docker

## Member repositories
- ray-project/kuberay (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:06.972537+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:18:53.411775+00:00, confidence not recorded.
  - readme: https://github.com/ray-project/kuberay (fetched 2026-08-28T04:07:06.972537+00:00, sha 041b4b4d550f)
- Data as of 2026-08-30T08:39:29.467469+00:00.