Description
NVIDIA is seeking a Deep Learning Performance Software Engineering Intern to support research and development of GPU-accelerated deep learning software. The role involves creating and maintaining SKILL, Wiki, and agent harnesses; developing TileGym, Triton CUDA TileIR backend, and CUDA Tile; creating highly optimized tile-based GPU kernels; and performing end-to-end performance optimization, analysis, and tuning. Candidates should be pursuing an engineering or computer science degree, with strong C/C++ and software design skills, and experience with agentic systems, GPU programming, performance modeling, profiling, debugging, and code optimization.
