Use new Metaflow Sandboxes to evaluate Metaflow and the full infrastructure stack behind it in the browser without having to install anything locally, like shown in the teaser video below.
Many of us have experienced how moving to a new house or apartment can improve the quality of life dramatically. However, there’s quite a bit of upfront cost involved in moving, especially if you have a large household. Before deciding to move to a new place, it is a good idea to tour the neighborhood for a few days to see if you like the environment.
It is a similar situation with any infrastructure investment: Effective infrastructure can drastically improve the quality of life for the whole organization but choosing and deploying a suitable stack takes some upfront investment. Metaflow is no exception.
While you can get started with Metaflow just by executing pip install metaflow on your laptop, to experience all its benefits, such as scalable compute, shared experiment tracking, and the Metaflow GUI (among many others), you need to deploy and configure backend services on your cloud account. Especially if you are not familiar with tools like Terraform and CloudFormation, deploying the full stack can take a while.
To make sure everyone can experience the full power of Metaflow effortlessly, we created new Metaflow Sandboxes! With a click of a button, you get a personal sandbox that you can use to test Metaflow and the full infrastructure stack behind it in the browser, without having to install anything locally.
Consider Metaflow Sandbox as AirBnB for data science and ML infrastructure: Sign up for one to explore the neighborhood for a few days, so you can decide if you are ready to move there permanently. Oh, and they are free!
The Full Stack Experience
Metaflow Sandboxes include the full stack of data science infrastructure. Each sandbox comes with:
- An auto-scaling Kubernetes cluster, which allows you to test scalable compute with Metaflow’s
@kubernetesdecorator. - A private S3-based datastore, built-in Metaflow.
- A metadata service for experiment tracking, which keeps track of all Metaflow executions happening inside the sandbox. You can query any past results using the Client API.
- Metaflow GUI, which allows you to monitor executions.
- In-browser VSCode editor with an embedded tutorial, so you can learn Metaflow and create flows without leaving the browser.
We will add new features in the sandboxes, like Argo Workflows for workflow orchestration, over time. Let us know on Slack what features you would like to see!
You can run any Metaflow flows in the sandbox, go through tutorials, and even access publicly available datasets in the sandbox. By default, you have a few days to use the sandbox. If you need more time, just let us know on Slack.
We hope that the sandbox convinces you to move to the Metaflow neighborhood and join hundreds of other forward-looking companies! Once you are done with your evaluation, you can deploy the same stack that’s available in the sandbox in your own cloud environment using our CloudFormation and Terraform templates. Or, if you want to equip your organization with sandbox-like data science environments with enterprise-level scale, security, and policy guarantees, schedule a call with us – we are happy to tell you more!