Description

You will develop and optimize CUDA Core Libraries in C++ and Python, owning features from design through maintenance.

This role is on-site.

Responsibilities

• Develop and implement parallel algorithms and idiomatic language bindings for core CUDA functionality.
• Optimize GPU algorithms and APIs, tuning for memory, parallelism, and synchronization.
• Own features end-to-end: development, testing, benchmarking, documentation, and long-term maintenance.
• Improve developer experience across the stack by enhancing CI, tests, packaging, and examples.
• Collaborate with senior engineers on design/code reviews and engage with users on performance investigations.

Required Skills

• 8+ years of related development experience.
• Strong programming skills in C++ and/or Python with a focus on systems-level software.
• Solid understanding of modern C++ (templates, generics, STL) and/or Python library development.
• Practical experience with parallel or heterogeneous programming (CUDA, OpenMP, GPU-accelerated Python).
• Experience contributing to production software or open-source libraries, including testing and profiling.
• Ability to navigate large, multi-language codebases (C++, Python, CMake, CI systems).
• BS, MS, or PhD in Computer Science, Computer Engineering, or equivalent experience.

Preferred Skills

• Hands-on experience with PyTorch, JAX, Numba, or CuPy.
• Familiarity with Thrust, CUB, libcudacxx, or compiler infrastructure (LLVM/Clang).
• Strong understanding of CPU/GPU architecture and hardware performance implications.

Key Skills

Education

Any Gradute