Our Blog

by Thomas Jardine, Head of Data Science - Americas @ BMLL

Case Study

BMLL Technologies processes petabytes of nanosecond-precision market data for the world's most sophisticated trading teams. In this case study, they share what a 48x performance improvement over pandas looks like in practice, and what it means for execution analysis at scale.

Tue, 25 Aug 2026

by Thijs Nieuwdorp

Tech

How to scope a pandas to Polars migration, from a single performance-sensitive section to the whole pipeline, and how to execute a full migration by hand or with an LLM.

Thu, 6 Aug 2026

by Thijs Nieuwdorp

Announcement

Polars Cloud 0.10.0 streams distributed query results into Python with sink_batches(), adds an experimental miso query planner, distributed pl.collect_all(), experimental HDFS support, optimized hive-partitioned scans, and a smarter distributed planner.

Tue, 4 Aug 2026

by Thijs Nieuwdorp

Tech

Explore a subset of Polymarket orderbook data locally, then run the exact same Polars query on 16 billion rows with Polars Cloud and serve it with Plotly Dash.

Tue, 28 Jul 2026

by Thijs Nieuwdorp

Announcement

Polars 1.43 adds pl.list() for packing columns into nested list values, ewm_sum/ewm_sum_by for exponentially weighted moving sums, faster joins on hive-partitioned data, a build_side join hint, and an O(n) rolling min_by/max_by, and more.

Thu, 23 Jul 2026

by Chiel Peters and Ritchie Vink

Tech

In this post we are comparing Polars on a single node vs in a distributed environment and highlight the differences.

Tue, 14 Jul 2026

by Thijs Nieuwdorp

In Aggregate

This edition covers Polars On-Prem on Kubernetes, the Polars vs Spark benchmark, releases 1.40 to 1.42, and optimizer improvements that make queries faster without any code changes.

Tue, 7 Jul 2026

by Thijs Nieuwdorp

Announcement

Polars Cloud 0.9.0 runs expressions distributed, ships large cloud I/O performance gains, a reworked ClusterContext API, a distributed Iceberg sink, and more.

Thu, 2 Jul 2026

by Thijs Nieuwdorp

Announcement

Polars 1.42 adds adaptive cloud I/O concurrency, optimizer elimination of contradictory filter predicates, and is_sorted() for DataFrame and Expr.

Mon, 29 Jun 2026

by Thijs Nieuwdorp

Tech

We had AI translate a pandas corpus to Polars. Most one-shot translations ran and the outputs matched. The code was idiomatic in most cases. Two structural patterns remain, and our new skill improves the model's Polars capability.

Thu, 25 Jun 2026

by Polars dev team

Tech

Polars is 3x faster than Spark and more than 6x faster on a single node, on the PDS-H benchmark.

Tue, 16 Jun 2026

by Chiel Peters

Announcement

We are launching our distributed engine on Kubernetes (e.g. on-prem, AKS & GKE). Try it for free. It also ships with advanced query profiling and OpenLineage support.

Wed, 3 Jun 2026

by Thijs Nieuwdorp

Announcement

Polars 1.41 brings a hand-written Thrift decoder for faster Parquet metadata reads, nested common subplan elimination in the optimizer, and LazyFrame.gather.

Tue, 26 May 2026

by Thijs Nieuwdorp

Tech

How Polars handles schema mismatches across CSV schema inference, multi-file Parquet datasets, Delta Lake writes, Apache Iceberg, and when you need to take control manually.

Thu, 30 Apr 2026

by Thijs Nieuwdorp

In Aggregate

This edition covers the streaming engine's expansion across sinks and joins, new Iceberg and Delta I/O, Polars Cloud query profiling, and the new PartitionBy API.

Thu, 16 Apr 2026

by Thijs Nieuwdorp

Tech

Learn how Polars streaming sort-merge join works, the concepts behind the algorithm, and how you can verify the query planner chose the merge-join.

Thu, 9 Apr 2026

by Thijs Nieuwdorp

Tech

Introducing the Polars Cloud query profiler. In this post we use it to identify performance bottlenecks and select the right instance family, instance type, and cluster size for your workload.

Thu, 26 Mar 2026

by Thijs Nieuwdorp

Tech

How to schedule and run Polars Cloud queries using Airflow, covering fire-and-forget jobs, parallel execution, multi-stage pipelines, and cluster management.

Tue, 17 Feb 2026

by Thijs Nieuwdorp

Tech

Exploring the evolution of the Categorical data type in Polars.

Thu, 29 Jan 2026

by Freerk Venhuizen, Senior Data Scientist @ Rabobank

Case Study

Window functions are a perfect fit for detecting periodic payment bookings at Rabobank. They not only improve performance but also makes code easier to read.

Wed, 10 Dec 2025

by Thijs Nieuwdorp

In Aggregate

This edition covers the launch of Polars Cloud, the new streaming engine, the Categoricals overhaul, the stabilization of Decimals, introduction of Int128, and the introduction of runtimes.

Wed, 3 Dec 2025

by Tyler Riccio, Assistant Vice President Enterprise Data and Analytics @ Citizens

Case Study

With Polars, Citizens achieved faster development cycles, richer and more timely insights, and created a strong pillar for the ongoing modernization journey.

Wed, 26 Nov 2025

by Dr. Michael Tomka - Actuary @ La Mobilière and Dr. Santiago Padrón - Senior Data and Machine Learning Consultant @ D ONE

Case Study

La Mobilière always keeps its promise. Now, even more promptly, thanks to Polars.

Thu, 9 Oct 2025

by Ritchie Vink

Announcement

We have raised €18M in Series A funding the build the next generation data.

Mon, 29 Sept 2025

by Arnaud Vennin (Tech Lead Data Engineer @ Decathlon)

Case Study

Decathlon adopts Polars to replace increasing number of workloads on Spark to significantly decrease infrastructure cost and complexity.

Mon, 15 Sept 2025

by Ritchie Vink

Announcement

We launch Polars Cloud; the distributed Cloud Architecture to run Polars anywhere

Wed, 3 Sept 2025

by Ritchie Vink

Talks

Ritchie Vink introduces Polars Cloud, the platform to scale Polars remotely with a novel distributed engine

Thu, 31 Jul 2025

by Thijs Nieuwdorp (Xomnia), Tom Drabas (NVIDIA)

Tech

With the introduction Unified Virtual Memory (UVM), larger than memory data processesing is possible with the GPU engine. Learn more about the implementation and performance considerations.

Thu, 26 Jun 2025

by Ritchie Vink

Tech

A fresh run of the PDS-H benchmark suite shows the performance of Polars versus other query processing engines.

Sun, 1 Jun 2025

by Dr. Patrick Bormann (Tech lead data scientist) @ METRO.digital

Case Study

METRO.digital migrated from SQL to Polars for KPI creation across 19 countries, improving code maintainability and reducing resource requirements while handling large datasets more efficiently.

Thu, 24 Apr 2025

by Orson Peters

Talks

Orson Peters walks through the core architecture and internals of the new Polars streaming engine

Wed, 9 Apr 2025

by Ritchie Vink

Announcement

We introduce Polars Cloud, the distributed cloud architecture to scale and run Polars anywhere.

Mon, 10 Feb 2025

by Rodrigo Girão Serrão

Tech

Explore the calendar-aware date operations that Polars supports to simplify working with dates.

Mon, 23 Dec 2024

by Rodrigo Girão Serrão

Tech

Polars has a rule that one expression must produce a single column as a result in a query. How do some expressions break this rule?

Thu, 5 Dec 2024

by Rodrigo Girão Serrão

In Aggregate

This edition covers plenty of performance improvements made to I/O, inequality joins, and much more.

Wed, 4 Dec 2024

by Rodrigo Girão Serrão

Tech

Learn about all Polars data types, including when to use them and how to use them.

Tue, 26 Nov 2024

by Ritchie Vink

Talks

Ritchie Vink presents Polars GPU acceleration at PyData NYC 2024

Mon, 25 Nov 2024

by Ritchie Vink

Talks

Ritchie Vink presents Polars 1.0 and the arrival of GPU acceleration at PyData Amsterdam 2024

Tue, 22 Oct 2024

by Sebastian Folz, Maren Westermann (Machine Learning Engineers @ DB Systel GmbH)

Case Study

The largest german railway company achieves nearly 20x speedup in data processing and identifies significant potential for cost savings on cloud resources.

Thu, 17 Oct 2024

by Liam Brannigan (Data Team Lead) @ Joulen

Case Study

Joulen builds a highly scalable and cost-effective infrastructure with Polars to power tools to optimize the use of renewable assets.

Wed, 25 Sept 2024

by Ritchie Vink, Robin van den Brink, Lawrence Mitchell

Tech

Polars and NVIDIA RAPIDS have worked together to bring GPU acceleration to Polars.

Tue, 17 Sept 2024

by Danny Correia (Software Engineering Manager) @ Vydia

Case Study

Vydia uses Polars' Ruby interface to ingest and process a fast growing number of CSVs files on their Ruby based platform that empowers music labels.

Fri, 13 Sept 2024

by Marco Gorelli (Polars core developer)

Tech

With Polars 1.6 a new lightweight plotting engine was introduced to create plots without additional dependencies.

Wed, 4 Sept 2024

by Ivano Malavolta and Karolina Bargieł

Tech

A research team of the VU Amsterdam published a paper benchmarking the energy usage and performance of pandas and Polars data analysis Python libraries.

Thu, 22 Aug 2024

by Isaac Robbins (Data Engineer) & Nelson Griffiths (Engineering & ML Lead) @ Double River Investments

Case Study

Double River achieves 10x speedups and 50% memory reduction in their financial models utilizing Polars plugins.

Wed, 14 Aug 2024

by Polars Development Team

Tech

Polars now supports native reading from Hugging face datasets

Fri, 9 Aug 2024

by Polars dev team

In Aggregate

This edition covers improved support for Hive partitioning, right joins, Hugging Face integration and much more.

Thu, 25 Jul 2024

by Ritchie Vink

Announcement

We have released Python Polars 1.0. We are very excited to reach this milestone.

Mon, 1 Jul 2024

by Jean-Philippe Dubuc, Senior Data Engineer @ MDPI

Case Study

MDPI, a pioneer in open access scientific publishing, uses Polars to process hundreds of GBs of textual data on their on-premise infrastructure on a daily basis.

Thu, 27 Jun 2024

by Robin van den Brink & Ritchie Vink

In Aggregate

This Polars in Aggregate shares more insights in some selected changes released in the versions Polars 0.20.17 to Polars 0.20.31.

Wed, 5 Jun 2024

by Ritchie Vink

Talks

Ritchie Vink delivers a keynote on Polars at PyCon Lithuania 2024

Thu, 2 May 2024

by Kieran Garvey (Senior L&D Partner) & Casey Haaland (Machine Learning Engineer) @ G-Research

Case Study

G-Research is a leading quantitative finance research firm that uses Polars in different projects throughout the organization.

Thu, 18 Apr 2024

by Stijn de Gooijer

Tech

A fresh run of the PDS-H benchmark suite shows the performance of Polars versus other query processing engines.

Tue, 16 Apr 2024

by Polars & NVIDIA

Tech

Polars and NVIDIA are bringing GPU acceleration to Polars dataframes in the near future. With the collaboration Polars is able to fully utilize both the CPU and GPU, significantly speeding up certain workloads.

Thu, 4 Apr 2024

by Polars dev team

In Aggregate

This edition covers support for streaming outer joins, zero-copy NumPy interop, COUNT(*) fast path, very fast boolean filters and ternary operations, and more.

Thu, 28 Mar 2024

by Chiel Peters

Tech

Predicate pushdown is one of the most important optimizations for a query engine. Read more about it in this article.

Tue, 19 Mar 2024

by Robin van den Brink, Paul Duvenage (Check Technologies)

Case Study

Check Technologies is a Dutch multi-modal mobility provider in the Netherlands. They have successfully migrated all data pipelines from pandas to Polars and saved 25% on their cloud bills.

Wed, 6 Mar 2024

by Robin van den Brink, Ritchie Vink

In Aggregate

Since the last issue of Polars in Aggregate, we had four Python releases (0.20.3 to 0.20.6) and a two Rust releases (0.36.2 and 0.37.0). These releases contained many performance improvements, bug fixes, breaking changes and new features.

Wed, 7 Feb 2024

by Ritchie Vink

Tech

We have completely rewritten the string/binary data structure we use in Polars.

Tue, 30 Jan 2024

by Chiel Peters

Tech

Read how Polars executes your query from planning, optimizing to execution.

Fri, 19 Jan 2024

by Ritchie Vink

Talks

Ritchie Vink introduces Polars at Global DevSlam 2023 in Dubai

Tue, 9 Jan 2024

by Robin van den Brink, Ritchie Vink

In Aggregate

Polars moves fast. We typically release once a week. In this blog we explain the most notable changes from the last month.

Thu, 4 Jan 2024

by Ritchie Vink

Talks

Ritchie Vink presents a keynote on the Polars plugin system, enabling users to extend Polars with custom functionality

Thu, 21 Dec 2023

by Ritchie Vink

Talks

Polars is an OLAP query engine built from scratch in Rust to fully utilize modern multi-core hardware

Wed, 29 Nov 2023

by Ritchie Vink

Talks

A deep dive into why the expression engine in Polars is so versatile and fast

Wed, 22 Nov 2023

by Stijn de Gooijer

Tech

An overview of breaking changes and deprecations in the latest Polars release.

Thu, 31 Aug 2023

by Ritchie Vink & Chiel Peters

Announcement

We are excited to announce the start of company that will build around Polars, enabling data processing at any scale.

Thu, 3 Aug 2023

by Ritchie Vink

Talks

Ritchie Vink explains what Polars does under the hood to make your data processing faster

Thu, 20 Jul 2023

by Ritchie Vink

Talks

Ritchie Vink presents how Polars has evolved beyond raw speed to tackle scaling challenges

Sat, 17 Jun 2023

by Polars

Tech

Benchmarking: under development

Sun, 1 Jan 2023

by Ritchie Vink

Talks

This talk will introduce Polars, a blazingly fast DataFrame library written in Rust on top of Apache Arrow

Sat, 22 Jan 2022

by Vincent Warmerdam

Tech

Guest post by Vincent D. Warmerdam, research advocate at Rasa, creator of the calmcode.io project and maintainer of many open source projects.

Sun, 21 Nov 2021

by Ritchie Vink

Talks

Bristech Bytesize #17: Polars, the fastest DataFrame library you never heard of... Panel with Joris van den Bossche & Jorge C. Leitão.

Thu, 16 Sept 2021

by Ritchie Vink

Talks

Ritchie Vink introduces Polars, the fastest DataFrame library you never heard of, at Xomnia's Data & Drinks series

Mon, 22 Mar 2021

by Ritchie Vink

Announcement

Some design decisions around the Polars DataFrame library.

Sun, 28 Feb 2021