Senior Software Engineer, CUDA Driver

What you'll do

• Architect and implement new features for the CUDA programming model and APIs.
• Develop high-performance, maintainable code for multiple operating systems.
• Build and maintain performance and precision models for GPU computing.
• Coordinate development efforts across multiple internal and external teams.
• Design software solutions for kernel mode components, compilers, and networking.
• Collaborate with hardware architecture teams to integrate hardware features into software.
• Solve complex system-level problems involving memory hierarchy, interconnects, and interrupts.

What we're looking for

• Bachelor of Science or Master of Science degree in Computer Science, Electrical Engineering, or a related field (or equivalent experience).
• 5+ years of relevant experience in developing systems software.
• Strong C programming skills.
• Experience designing, debugging, and maintaining complex software stacks.
• Experience with operating system interfaces for threads, process control, and virtual memory.
• Experience with HW/SW co-design, performance modeling using emulation/simulation, and developing SW programming model exposures for HW features.
• Understanding of system-level architecture, including interconnects, memory hierarchy, interrupts, and memory-mapped IO.
• Strong interpersonal, verbal, and written communication skills to achieve objectives under tight time constraints.
• Experience with kernel scheduling, task runtimes, kernel-mode development, and Linux systems software (preferred).
• Strong background in parallel computing, preferably writing CUDA programs or CUDA-based libraries (preferred).
• Knowledge of memory coherence and consistency models in concurrent/parallel systems (preferred).
• Experience maintaining and extending programming models or higher-level language support for Linux or similar environments (preferred).
• Familiarity with distributed training/inference patterns and deep learning frameworks (preferred).