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:
- 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)
- Kubernetes (minikube, on-premise, AWS EKS or GKE or any public cloud)
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
- 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
A comprehensive documentation is available here link
If you have a question or are looking for guidance, we recommend opening an issue so a member of the community can help!
Please write to us.
Amol Umbarkar (email: [email protected] / twitter)