Introduction
The Groundzero DuckDB runtime runs your Python code next to the DuckDB analytics engine. It is the fastest way to join, filter and aggregate data that fits on one machine. You read from any Groundzero connection into DuckDB relations, combine them with SQL, and either return the result (datasets, previews) or write it back to storage and databases (jobs).
This guide shows how to run your first job, dataset and notebook, how to read and write data, and how to use the Groundzero helper functions. It also covers the installed packages, limits, troubleshooting and a gallery of runnable examples.
Who this guide is for
Developers writing datasets, ETL jobs and notebooks on Groundzero. You
should be comfortable with Python and SQL. All examples use fictional
data and connection names such as sales_s3 and orders_pg; replace them
with the names of your own connections.