From Python Code to Containers : Packaging with uv and Docker

Python code can run on your laptop and still fail when a colleague runs it or a scheduled job starts. Differences in Python versions, dependencies, and system libraries can cause these failures.

Join PyData Chiang Mai for a practical talk on packaging Python data pipelines and managing their environments with uv and Docker.

We will follow one small Python application from local development to a container. We will define the project in `pyproject.toml`, manage its environment with uv, and build a Docker image from its locked dependencies.

We will also apply these ideas to a data pipeline that runs dbt Core through Apache Airflow with Astronomer Cosmos. This example will show how to separate dbt’s Python dependencies from Airflow’s environment and when to use a virtual environment or a container.

You will leave with a practical workflow for packaging a Python data pipeline and choosing where to isolate its dependencies.