Senior Research Engineer - CUDA AI

- Design and construct evaluation frameworks to assess AI models proficiency in generating, optimizing, and maintaining CUDA code throughout the entire software development lifecycle - Develop benchmarks that accurately reflect real-world CUDA programming patterns and use cases within NVIDIAs ecosystem (including kernels, libraries, and multi-GPU applications) - Contribute to avant-garde AI research initiatives such as innovative training methodologies, tool development, and dataset curation - Collaborate with teams creating CUDA-oriented AI tools to share evaluation insights, pinpoint performance gaps, and incorporate novel features (for example, RAG, profiling, web research) - Create and maintain high-quality datasets, utilizing both synthetic generation and real-world CUDA code to enhance AI-powered programming - Investigate and devise new AI tools for developers, including IDE enhancements, cloud-based profiling services, and ergonomic interfaces for agents - Conduct experiments to validate new strategies in areas such as reinforcement learning for code optimization and multimodal representation learning - Lead initiatives to broaden our teams influence across various CUDA application domains and levels of complexity