🏃‍♀️Runhouse Overview🏠

Programmable remote compute and data across environments and users

Join Discord

Runhouse is a unified interface into existing compute and data systems, built to reclaim the 50-75% of ML practitioners’ time lost to debugging, adapting, or repackaging code for different environments.

Runhouse lets you:

  • Send functions, classes, and data to any of your compute or data infra, all in Python, and continue to interact with them eagerly (there’s no DAG) from your existing code and environment.

  • Share live and versioned resources across environments or teams, providing a unified layer for accessibility, visibility, and management across all your infra and providers.

It wraps industry-standard tooling like Ray and the Cloud SDKs (boto, gsutil, etc. via SkyPilot to give you production-quality features like queuing, distributed, async, logging, low latency, hardware efficiency, auto-launching, and auto-termination out of the box.

Who is this for?

  • 🦸‍♀️ OSS maintainers who want to improve the accessibility, reproducibility, and reach of their code, without having to build support or examples for every cloud or compute system (e.g. Kubernetes) one by one. See this in action in 🤗 Hugging Face Transformers, Accelerate and 🦜🔗 Langchain.

  • 👩‍🔬 ML Researchers and Data Scientists who don’t want to spend or wait 3-6 months translating and packaging their work for production.

  • 👩‍🏭 ML Engineers who want to be able to update and improve production services, pipelines, and artifacts with a Pythonic, debuggable devX.

  • 👩‍🔧 ML Platform teams who want a versioned, shared, maintainable stack of services and data artifacts that research and production pipelines both depend on.

Table of Contents

Contributing and Community