Get Started with RAPIDS

Learn how to use our libraries, accelerate common data science use cases, and access hands-on labs to develop business-critical applications.


Download Now  Join Community

Explore Resources to Get Started With RAPIDS Libraries

Get started with RAPIDS suite of software libraries

cuDF is a GPU DataFrame library that accelerates pandas workflows on GPUs with zero code changes. cuDF is available by default on Colab.



Polars provides a GPU engine powered by cuDF that delivers up to 13X speedup over CPU, allowing users to process hundreds of millions of records in seconds.



With the NetworkX GPU backend, you can accelerate popular graph algorithms with zero code changes with cuGraph, a GPU-accelerated graph analytics library.



cuML is a unified CPU/GPU library for executing machine learning (ML) algorithms with an API that closely follows the scikit-learn API.


Use Case Tutorials

Learn how to accelerate data science in a few short hours in simple technical walk-throughs that pair step-by-step tutorials with notebooks, sample datasets, and benchmarks in example exercises that demonstrate how RAPIDS helps data scientists save time and better leverage their data in ubiquitous data science and machine learning use cases.

Accelerated Data Analysis

Learn how to use RAPIDS cuDF to accelerate data analytics workflows.


This starter kit covers how to use RAPIDS for:


  • Exploratory Data Analysis
  • Time Series Analysis
  • Visualization
  • DataFrame to ML

Get Started

Data Processing at Scale

Explore how to accelerate Apache Spark 3.0 workloads with the RAPIDS Accelerator for Apache Spark.


This starter kit walks through an example use case in retail:


  • Data Processing, including ETL and manipulation functions

Get Started

Time Series Forecasting

Build time series forecasting models with RAPIDS cuDF and cuML, which mimic pandas and sci-kit learn.


This starter kit includes two examples of time series workflows:


  • Time Series Analysis

Get Started

Learn How Top-Performing Kaggle Grandmasters Win Data Science Competitions

The Kaggle Grandmasters of NVIDIA place in the top 10% of Kaggle Grandmasters worldwide. Learn their tricks of the trade with our Kaggle Grandmaster series that explain how our teams use accelerated data science to create winning recommender systems and predictive models.

Watch video about mastering multilingual recommender systems

Mastering Multilingual Recommender Systems

The KGMoN leveraged transfer learning, model embedding, and other innovative techniques to construct a two-stage pipeline solution for e-commerce product suggestions. This solution won the 2023 KDD Cup.

View Video Read Blog
Watch video about mastering recommender systems

Mastering Recommender Systems

KGMoN tackled an e-commerce use case by training high-functioning recommendation systems with a dataset that included millions of anonymized user interactions. These strategies placed top 3 in Kaggle’s Otto competition.

View Video Read Blog
Read blog about predicting credit defaults using time-series models

Predicting Credit Defaults on a Massive Dataset

Jiwei Liu, KGMON, walks through GPU-acceleration in a default prediction use case (top 10 in Kaggle’s American Express competition) that resulted in significantly faster inference times.

Read Blog

Join Our Community

Decorative icon representing NVIDIA Slack community

Join Our Community on Slack

Decorative icon representing NVIDIA Developer Forums

Developer Forums

Decorative icon representing NVIDIA newsletter subscription

Subscribe to our newsletter

RAPIDS libraries are open source, written in Python, and built on Apache Arrow. The software is being developed in partnership with enterprises globally.
Download RAPIDS to dramatically accelerate machine learning and data science.


Download Now