Metaflow sandboxes are now loaded with many new interactive ML tutorials: You can learn computer vision, natural language processing, and recommendation systems, and much more right in your browser, backed by a full stack of ML infrastructure, powered by Metaflow.
Metaflow sandboxes make it easy to explore modern ML infrastructure conveniently in the browser, for free, powered by open-source Metaflow. You can use your personal sandbox to get a feel for real-world ML and the infrastructure behind it, including foundational features like data access, compute, orchestration, versioning, and modeling.
Today, we are releasing a major update to our sandboxes: We have loaded them with a growing library of curated, interactive environments which teach you computer vision, natural language processing, recommendations systems, deep learning, and more. The new workspaces feature production-grade Metaflow flows, teaching you patterns that you can use to build real-world applications, going beyond the usual notebook examples.
- It is important to learn not just the models powering ML applications, but also the infrastructure and workflows (human and technical) around them – everything related to MLOps. Inevitably, you will need to face these concerns when building real-world applications.
- Setting up ML infrastructure requires engineering effort and access to a cloud account, which is not always available to individuals. Sandboxes allow you to test the full Metaflow stack for free without having to go through a chain of approvals and cloud-related hurdles.
Consider the new workspaces as starting points for your prototypes and demos. Beyond pre-built content in the sandbox, you can use the workspaces to experiment as you see fit. After all, data science and ML are all about experimentation and iteration!
If you’re looking to deploy a recommender system, building upon a foundation model, or you are working on a more traditional business data science project, the sandboxes provide an ideal platform for testing out ideas in a realistic environment.
Sandboxes are not a toy environment. Your experiments run on a production-grade infrastructure stack, leveraging common ML libraries, cloud-based data, Kubernetes-based compute and workflow orchestration. Once you are ready to move your experiments to production, you can deploy the infrastructure stack in your environment, either through open-source templates or through our managed service, Outerbounds Platform, both of which run on your cloud account.
Get started today
It is easy to get started: All you need is a Google or GitHub account. The sandboxes are fully browser-based, including a cloud workstation with a built-in VSCode, pre-installed with your personal, cloud-based Metaflow stack. Pick one of our existing tutorials or other workspaces and start learning and exploring! Also, join us and over 2000 other data scientists and engineers on Slack for support and to share ideas for future content.