Evaluate Metaflow for free, right from your Browser
Get StartedMetaflow Sandbox allows you to experiment with modern ML and the infrastructure behind it, powered by open-source Metaflow, without having to install anything locally.
Click one of the workspaces below to start exploring!
Some workspaces for you to explore Metaflow
- Open in Sandbox
Welcome to Metaflow Sandbox
Get familiar with the features and functionality contained in your sandbox.
- Open in Sandbox Open Tutorial
Metaflow Tutorial
Learn the fundamentals of using Metaflow to power machine learning workflows.
- Open in Sandbox Read Blog
Audio Processing with Whisper
Use OpenAI's Whisper model in a production-grade workflow by applying MLOps practices with Metaflow.
- Open in Sandbox Open Tutorial
Natural Language Processing
Build, train, test, and deploy a machine learning model that performs text classification.
- Open in Sandbox Open Tutorial
Computer Vision - Beginner
Use Keras to train and evaluate a machine learning model that performs image classification.
- Open in Sandbox Open Tutorial
Computer Vision - Intermediate
Use PyTorch and Metaflow to create scalable image processing and deep learning workflows.
- Open in Sandbox Open Tutorial
Full stack Recommender System
Use popular open-source libraries and tools including DuckDB, Gensim, Metaflow, and Keras to build a fully working cloud endpoint that serves predictions in real-time.
Frequently asked questions
What is Metaflow sandbox exactly?
Metaflow Sandbox is a free, cloud-based development environment for ML and data-intensive applications, powered by open-source Metaflow. Behind the browser-based VSCode interface, it includes a scalable Kubernetes-based compute cluster, a Metaflow UI, an S3-based datastore, a metadata service, and workflow orchestration systems. In other words, everything you need to develop modern ML applications with your favorite libraries!
What can I do with a Metaflow Sandbox?
Use your personal sandbox to evaluate all features of Metaflow, including scaling flows with @kubernetes, notebook integration, and visualization using @cards. As you have access to the full-stack of ML infrastructure, you can use sandboxes to learn how to build modern ML applications, as instructed by our tutorials and articles, linked above.
Can I use a Metaflow Sandbox for production use cases?
Sandboxes are not meant for production use cases. To use Metaflow at work, deploy it in your own cloud account either by using our open-source templates or by adopting Outerbounds Platform.
How long may I use the Sandbox?
Sandboxes are automatically terminated after a few days of inactivity, after which all data and code is lost. If you want to work with your sandbox for a longer time, send us a note on the #ask-metaflow channel on our Slack and we are happy to extend the time for you. Feel free to request a new sandbox if your existing one has been terminated.
I have a question that is not answered here
Join our Slack and ask your question on #ask-metaflow - there are no silly questions! We will help you there promptly. Alternatively, you can email hello@outerbounds.co.
Metaflow at Work
If you enjoyed the development experience with Metaflow, backed by managed compute, versioning, workflow orchestration, take a look at Outerbounds Platform which provides the same benefits and more, deployed securely in your cloud account.
Learn about the Outerbounds Platform