Krayne

CLI and SDK for creating, managing, and scaling Ray clusters on Kubernetes.

Krayne wraps the KubeRay operator behind a clean, opinionated interface so ML practitioners can get distributed compute without touching Kubernetes manifests.

---

Get started in seconds

pip install krayne
krayne create my-cluster --gpus-per-worker 1 --workers 2

That's it. One command gives you a fully configured Ray cluster with dashboard, notebook, and SSH access.

---

Interactive TUI

A fast and intuitive terminal UI is also available:

ikrayne

Navigate clusters, create with prefilled forms, scale, delete, and toggle tunnels — all with keyboard shortcuts.

---

Why Krayne?

Pain point	How Krayne helps
Verbose KubeRay YAML manifests	krayne create my-cluster — zero config needed
Kubernetes expertise required	Sensible defaults handle resources, services, and networking
No automation path	Python SDK exposes the same operations as functions
Scattered tooling	One CLI for create, scale, describe, delete, and tunnels

---

CLI and Python SDK

CLIPython SDK

krayne create my-experiment --gpus-per-worker 1 --workers 2
krayne describe my-experiment
krayne scale my-experiment --replicas 4
krayne delete my-experiment --force

from krayne.api import create_cluster, scale_cluster, delete_cluster
from krayne.config import ClusterConfig, WorkerGroupConfig

config = ClusterConfig(
    name="my-experiment",
    worker_groups=[WorkerGroupConfig(replicas=2, gpus=1, gpu_type="a100")],
)

info = create_cluster(config, wait=True)
print(f"Dashboard: {info.dashboard_url}")

scale_cluster("my-experiment", "default", "worker", replicas=4)
delete_cluster("my-experiment", "default")

---

Key features

• Zero-config defaults — every command works with no flags
• CLI and SDK — anything from the terminal works from code too
• Interactive TUI — k9s-style keyboard-driven cluster management
• GPU support — one flag to add GPUs with type selection
• Local sandbox — krayne sandbox setup for development without a real cluster
• Pydantic config — validated configuration with YAML override support
• Rich output — beautiful terminal tables, with --output json for scripting

---

• Quickstart

  ---

  Install Krayne and create your first cluster in under 5 minutes.

  Get started

• Core Concepts

  ---

  Understand Ray clusters, KubeRay, and the cluster lifecycle.

  Learn more

• CLI Reference

  ---

  Full reference for every krayne command, flag, and option.

  CLI docs

• Python SDK

  ---

  Use Krayne programmatically in scripts, notebooks, and pipelines.

  SDK docs