A Python data pipeline and notebook project on Polars and DuckDB, managed with uv and make. Tested transforms turn raw CSV into Parquet with explicit schemas and a rejects file, DuckDB queries the result, and marimo notebooks lint, typecheck and run in verify.
An interactive Python data app on Streamlit, managed with uv and make. Sidebar filters drive a metric, an Altair chart and a table over a local CSV; the data logic is a tested Polars package, AppTest drives the page, and verify probes the headless server.