Airflow Beyond the Cloud: Python Workflows at the Edge

Constance Martineau

Apache Airflow is famous for orchestrating massive data pipelines, but what if you could also use it to orchestrate things in your physical workspace?

In this talk, we'll explore how Airflow 3's new lightweight architecture makes micro-orchestration possible on resource-constrained devices. I'll demonstrate a real working system where an Airflow Edge Worker running on a Raspberry Pi controls an "On Air" LED that automatically lights up when a Zoom call starts. All built with standard Python and familiar Airflow patterns.

What starts as a desk toy points to a bigger trend: orchestration breaking free from the data centre and moving closer to where work actually happens. You'll walk away seeing Airflow (and Python orchestration in general) not just as a data-engineering tool, but as a bridge between the digital and physical worlds.

Basic familiarity with Python is all you need, no prior Airflow experience required.

About Constance Martineau

Constance Martineau is a Staff Product Manager at Astronomer and a long-time contributor to the Apache Airflow open-source project, where she’s helped shape features across Airflow 3.

She works at the intersection of data engineering, orchestration, and developer experience — turning complex infrastructure into tools that feel intuitive and fun. When she’s not working on new features, she’s usually automating something else, like her coffee machine, her lights, or (in this case) an “On Air” sign that knows when she’s in a meeting.

Constance lives in Montréal and spends her free time painting, experimenting with Python side projects, and trying (unsuccessfully) to keep her plants alive.