hyperML

Radically simple, efficient platform to abstract infrastructure complexities from data scientists and ML engineers. The goal of the project is to integrate your favourite IDE (Jupyter, VS Code) with cloud services in the backend.

hyperML works on top of kubernetes to provide reusable environments, launch or schedule notebooks and python jobs. If you are sharing resources then you can queue up requests as well.

Demo Links:

Launch Notebook
CLI
Schedule Notebooks from Jupyter Labs

Problems solved

Ease of use (follows general dialect of data science)
Abstracts infrastructure from data scientists and ML engineers leting them focus more on science
Extend your local environment with on-demand cloud resources without having to leave your favourite IDE (jupyter labs at the moment but VS code extension is planned).
Scale ML experiments by effortlessly launching new notebooks or scheduling them to run in background
Share infrastructure resources especially when there is shortage of it
Hassle free environments through container images
Deploy models to containers with a single command (coming soon)

Requirements

Kubernetes (minikube, on-premise, AWS EKS or GKE or any public cloud)

Getting Started

Install standalone binary

curl -LO curl http://storage.googleapis.com/hyperml/releases/0.9.0/hyperml /usr/local/bin/hyperml

You can also install hyperML as lambada function to optimize server costs.

A host of quick start configuration guides are available on hyperML website

Local install
Minikube Install
Kubernetes Install
AWS EKS
GCP GKE

User flow:

Run or Schedule Notebooks to run in the background right from the comform of jupyter labs
Launch notebooks on click of a button
Install locally or on remote server (VM) or as a container inside kubernetes
Simple CLI to run Python code bundles inside containers on kubernetes

Documentation

A comprehensive documentation is available here link

Problems?

If you have a question or are looking for guidance, we recommend opening an issue so a member of the community can help!

Professional Support

Please write to us.

Author

Amol Umbarkar (email: amol@hyperml.com / twitter)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!