Senior Software Engineer, CUDA C++ Core Libraries
Core
Design and optimize high-performance CUDA C++ core libraries, parallel algorithms, and runtime abstractions to enable fast GPU-accelerated software for developers and AI agents.
Role type
Senior IC CUDA C++ Core Libraries Engineer
Builds
Foundational CUDA C++ libraries, parallel algorithms, and runtime infrastructure
Domain
HPC and AI, GPU computing, Systems programming
Deliverable
production ML models | product features
Required skills
Modern C++ (templates, type systems), Generic programming, Systems-level software development, CUDA or heterogeneous programming, API/ABI compatibility design, Performance optimization, Build systems and CI/CD
Preferred skills
GPU architecture knowledge, Thrust/CUB/libcudacxx, Compiler infrastructure (LLVM/Clang/MLIR), Python/Rust interoperability
Technologies
CUDA, C++, C, Python, Rust, Thrust, CUB, libcudacxx, LLVM, Clang, MLIR
Responsibilities
Design and implement foundational CUDA C++ libraries and parallel algorithms; Optimize GPU algorithms from high-level interfaces to low-level implementations; Design interoperability boundaries for C/C++ consumption from Python and Rust; Balance performance, portability, and API evolution; Own features end-to-end including testing, profiling, and documentation; Improve developer productivity via diagnostics and build integration; Collaborate on architecture and code reviews; Engage with users on performance and correctness issues
Seniority
Senior, hands-on IC