GPU Data Science with NVIDIA RAPIDS

• Familiarity with CUDA
• Python programming experience

Audience

• Data scientists
• Developers

RAPIDS is a suite of open source software libraries built to accelerate GPU-driven data science and analytics pipelines. It is based on Python and includes a DataFrame API that integrates with a variety of machine learning algorithms.

This instructor-led, live training (online or onsite) is aimed at data scientists and developers who wish to use RAPIDS to build GPU-accelerated data pipelines, workflows, and visualizations, applying machine learning algorithms, such as XGBoost, cuML, etc.

By the end of this training, participants will be able to:

• Set up the necessary development environment to build data models with NVIDIA RAPIDS.
• Understand the features, components, and advantages of RAPIDS.
• Leverage GPUs to accelerate end-to-end data and analytics pipelines.
• Implement GPU-accelerated data preparation and ETL with cuDF and Apache Arrow.
• Learn how to perform machine learning tasks with XGBoost and cuML algorithms.
• Build data visualizations and execute graph analysis with cuXfilter and cuGraph.

Format of the Course

• Interactive lecture and discussion.
• Lots of exercises and practice.
• Hands-on implementation in a live-lab environment.

Course Customization Options

• To request a customized training for this course, please contact us to arrange.

Introduction

• Overview of RAPIDS features and components
• GPU computing concepts

Getting Started

• Installing RAPIDS
• cuDF, cUML, and Dask
• Primitives, algorithms, and APIs

Managing and Training Data

• Data preparation and ETL
• Creating a training set using XGBoost
• Testing the training model
• Working with CuPy array
• Using Apache Arrow data frames

Visualizing and Deploying Models

• Graph analysis with cuGraph
• Implementing Multi-GPU with Dask
• Creating an interactive dashboard with cuXfilter
• Inference and prediction examples

Troubleshooting

Summary and Next Steps