Senior C++/CUDA Engineer for Open-Source GPU Data

NVIDIA Corporation; Santa Clara, Northern (CA, KY); Hybrid; USD 184,000 - 357,000; Full time

Benefits offered by this job

Equity

Comprehensive benefits

Job summary

NVIDIAs Analytics and Data Intelligence (ADI) group is building libcudf — an open-source CUDA C++ library that accelerates database and DataFrame operations for diverse domains like business intelligence, genomics, and LLM training. We seek an outstanding engineer and/or scientist to apply parallel programming to accelerate open-source libraries for GPU data processing.

You will drive speed-of-light performance from single workstations to multi-node GPU supercomputers, and build the

Qualifications

• 8+ years of experience in Computer Science or Software Engineering
• Strong Modern C++ programming skills
• Robust, readable, high-performance code
• Familiarity with RAPIDS cuDF
• Experience in distributed workflow development and debugging
• Open-source software development and publishing your work in technical blogs and conferences

Responsibilities

• Own development for UcxExchange in Velox.
• GPU-to-GPU communication in Presto: https://github.com/facebookincubator/velox/tree/main/velox/experimental/ucx-exchange
• Drive the feature roadmap, improve performance, and ensure correctness
• Optimize multi-node performance for analytical queries with Presto GPU
• Lead projects of interest in Presto, Velox, cuDF
• Such as operator completeness, GPU memory oversubscription, efficient CPU fallback
• Plus your ideas

Skills

Modern C++

Distributed systems

GPU CUDA

Open-source contribution

Education

MS/PhD in CS/Engineering

Tools

UCX

NCCL

Velox

Presto

cuDF

Job description

NVIDIAs Analytics and Data Intelligence (ADI) group is building libcudf — an open-source CUDA C++ library that accelerates database and DataFrame operations for diverse domains like business intelligence, genomics, and LLM training. We seek an outstanding engineer and/or scientist to apply parallel programming to accelerate open-source libraries for GPU data processing.

You will drive speed-of-light performance from single workstations to multi-node GPU supercomputers, and build the