Advanced Data Processing: Join Arrow Tables In-Memory (No Pandas) with PyArrow

You can join tables without Pandas, Polars, or materializing data into Python objects. In this video, we pull two datasets into Arrow Tables (think: a `users` table and an `orders` table), then use **PyArrow Compute** to join them *directly in-memory* using classic join types: - Inner join - Left outer join - Right outer join - Full outer join We also cover how to handle conflicting column names with left and right suffixes (so you can keep `status_user` and `status_order` straight), and how to chain operations like join + filter entirely in Arrow memory. If you are building data apps (including Streamlit), this is a clean way to bring data in via ADBC, join it fast, and render results without ever converting to a DataFrame. Chapters 0:00 Intro 0:30 Why in-memory Arrow joins are awesome 1:25 Setup: two simple tables (users + orders) 2:45 Inner, left, right, and full joins 3:35 Column suffixes for conflicting names 4:50 Join + filter chaining in Arrow Video Links: PyArrow Compute docs: https://arrow.apache.org/docs/python/compute.html ADBC quick starts: https://github.com/columnar-tech/adbc-quickstarts My Links: My bespoke devrel company: https://www.earlysignal.tech/ Substack: https://thefulldatastack.substack.com/ LinkedIn: https://www.linkedin.com/in/hoytemerson/ Like and subscribe for more Arrow + data stack deep dives.