Talk
Narwhals: One dataframe API to ~rule~ run them all
Thursday, May 28
15:20 - 16:05
RoomGnocchi
LanguageEnglish
Audience levelIntermediate
Elevator pitchAbstractTagsAnalytics, Data Engineering, Dataframes
What if your library didn’t care whether the input was pandas, Polars, PyArrow, or a DuckDB connection? With Narwhals it doesn’t. Same code, every backend (and SQL).
Python’s dataframe world is no longer just pandas: Polars, PyArrow, DuckDB, Modin, cuDF and Dask all coexist, and your users mix them.
Narwhals is a lightweight compatibility layer that lets library authors write code once and run it on every major backend, SQL engines included. Live on stage: one function, many backends; lazy-execution gotchas; Polars-style Python compiled into real DuckDB SQL, and more.
