A Kubernetes-native operator for managing development servers, particularly designed for PyTorch and ML development workloads. This project provides both a Kubernetes operator and a CLI tool for creating, managing, and accessing development environments.

📍 Current Status: This project is under active development.

• Kubernetes-Native: Built using the Kopf framework for Python.
• Resource Templates: DevServerFlavor CRDs define t-shirt sized resource configurations.
• User Management: DevServerUser CRDs for managing user access and SSH keys.
• Lifecycle Management: Automatic shutdown and expiration.
• CLI Tool: devctl for easy interaction with DevServers.
• SSH Agent Forwarding: Seamlessly forward your SSH agent to the DevServer.
• YAML Configuration: Configure devctl using ~/.config/devctl/config.yaml.
• Test-Driven Development: Comprehensive test suite using pytest and k3d.

• Python 3.9+
• Docker
• uv
• k3d
• kubectl
• A Kubernetes cluster (local k3d or remote)

# Create a local k3d cluster
k3d cluster create devserver-cluster

# Apply the Custom Resource Definitions and Flavors
make install-crds

# Start the operator in development mode
make run

In a new terminal:

# Create a development server
uv run devctl create --name mydev --flavor cpu-small

# List your servers
uv run devctl list

# Add a user
uv run devctl user add --name test-user --public-key-file ~/.ssh/id_rsa.pub

# Create a server with GPU support (see dev/eks/README.md for setup)
uv run devctl create --name my-gpu-dev --flavor gpu-small --image fedora:latest

# Delete a server
uv run devctl delete mydev

For more detailed documentation, please see the following README.md files:

• Source Code & Architecture
• Operator & CRDs
• CLI Reference
• Testing

Contributions are welcome! Please see the PROJECT.md for the development plan.

Before submitting a pull request, please run the pre-commit checks to ensure your changes adhere to the project's coding standards:

make pre-commit

1. Fork the repository.
2. Create a feature branch.
3. Make your changes with tests.
4. Run the test suite: make test
5. Submit a pull request.

This project is licensed under the Apache License, Version 2.0. See the LICENSE file for details.