Software Engineer, cuDNN - Deep Learning
Core
Design, build, and ship cuDNN, a GPU-accelerated library of primitives for deep neural networks, including optimized large language model (LLM) support.
Role type
Senior IC software engineer (GPU-accelerated deep learning primitives)
Builds
NVIDIA's AI software stack (cuDNN library)
Domain
AI / Deep Learning / GPU Computing
Deliverable
production ML models
Required skills
C/C++ development, CUDA development, Python, linear algebra, high-level software architecture design, performance analysis, profiling, code optimization
Preferred skills
GPU programming and optimization expertise (CUDA/OpenCL), practical experience with machine learning (deep learning), computer architecture knowledge, MLIR development, compiler optimization
Technologies
CUDA, Python, C/C++, cuDNN, LLMs, MLIR
Responsibilities
Develop production-quality software for NVIDIA's AI software stack; Analyze performance of workloads and propose software improvements; Collaborate with deep learning engineers and GPU architects on applications like generative AI and autonomous driving; Contribute to API design, software architecture, performance modeling, testing, and GPU kernel development.
