Leaderboard/Developer Tooling/skills
Last commit on August 3, 2026·Created on February 25, 2026

NVIDIA/skills

A technical library for accelerating data pipelines and optimization workflows on NVIDIA GPUs
Combined rank
#93
across all skills
In Developer Tooling
#5
category rank
Stars
2.8k
+2.8% in last 7d
Forks
324
+2.5% in last 7d
Watchers
18
-5.3% in last 7d
Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of real attention, adoption, and active maintenance.
Description

This repository provides specialized guidance for developers and data engineers working with GPU-accelerated computing. It focuses on transitioning heavy workloads—such as ETL processes, joins, and Parquet/CSV pipelines—from CPUs to GPUs using cuDF and dask-cuDF.

The collection also supports quant developers solving complex optimization problems, including linear and quadratic programming, routing, and Mean-CVaR portfolio optimization via cuOpt and cuFOLIO.

WHO IT'S FOR
Data Engineers
accelerate GPU data pipelines
Quant Developers
solve optimization portfolio problems
ML Training Engineers
run distributed RLHF training
Robotics Engineers
build sensor inference workflows
30 Days of GitHub Stars
+511 stars
↑ 23.4% monthly change
Jul 7Jul 14Jul 21Jul 28Aug 5
Compatibility & Install
327 Skills · 11 Compatible agents · 4 categories · 5 install paths
Compatible agents
Claude CodeCodexCursorGemini CLIOpenClawWindsurfAmpOpenCodeKiroHermesDeepSeek
Categories
Developer ToolingAutomationScienceSoftware Engineering
Also in this category
4 of 35