Skip to main content

Developing with Metaflow

Every project is a journey. Each stage of the journey presents different challenges and requirements. Luckily Metaflow allows you to start experimenting easily without having to worry about all details upfront. Rest assured that when your project grows in scale and impact, Metaflow helps your code grow accordingly.

A typical project goes through three major stages that are illustrated below. This section focuses on the first stage - local development. If you are curious about the next two stages, you can take a peek at Scalable Flows and Production Deployments.

Developing on a Personal Workstation

As a data scientist or engineer, your main productivity tool is your personal workstation, such as a laptop or a cloud workstation. A great thing about a personal computer is that they allow you to iterate quickly and experiment freely knowing that you are not accidentally interfering with a colleague's work.

Metaflow treats local development as the first class concern. You can develop and test Metaflow code locally like any other Python project or a notebook. Here's what often happens in the early phases of a project:

  1. Many data scientists are familiar with notebooks that shine at open-ended exploration and quick sketching of solutions. When developing with Metaflow, it is totally ok (although not required) to use notebooks for analysis. You can execute flows in a notebook and use the Metaflow Client API to inspect the results.

  2. Once you have a rough idea for the first version of your project, it is useful to structure it as a workflow, or flows in Metaflow parlance. Metaflow makes this easy: You can copy-paste the best parts of a notebook as steps of a Metaflow flow. For details why this is a good idea and how to create flows in practice, see Creating Flows.

  3. Don't consider flows just as static configuration. They are living and dynamic entities that you should be able to execute locally and improve gradually (this is where resume comes in handy!). The workflow becomes the backbone of your application - in particular helping with data flow through artifacts - which enables much of the functionality in the next phases of the project.

What You Will Learn

Let's go ahead and learn how to create and test Metaflow flows. This stage covers five core topics:

  1. Creating flows
  2. Inspecting results of flows
  3. Managing flows programmatically
  4. Visualizing results
  5. Debugging flows

These topics work locally on your workstation without any additional infrastructure, so it is easy to get started.