

#197: Polars with Jeroen Janssens and Thijs Nieuwdorp
Jul 23, 2025
In this engaging discussion, Jeroen Janssens, a developer relations engineer at Posit, and Thijs Nieuwdorp, a data scientist at Xomnia, explore the revolutionary power of the Polars library in Python data processing. They share insights on its impressive speed boosts, thanks to GPU collaboration, and why it outperforms traditional tools like Pandas. The duo also discusses their journey co-authoring 'Python Polars: The Definitive Guide,' including the challenges of technical writing and the significance of understanding user needs.
AI Snips
Chapters
Books
Transcript
Episode notes
Polars Journey from 30s to 1s
- Thijs improved an ETL pipeline runtime from 30 seconds to one second using Polars at a client.
- This achievement convinced a hesitant team to adopt Polars despite it being pre-1.0 version.
Polars API Enhances Readability
- Polars combines speed with a declarative API focused on expressions that build readable data workflows.
- This design helps newcomers better understand data transformations compared to imperative Pandas operations.
Rust and Lazy Execution Speedup
- Polars achieves speed by combining Rust's multithreading with a lazy execution model.
- Lazy evaluation lets Polars optimize data queries by applying filters early and loading only needed columns.