Senior Software Engineer, CUDA Core Libraries

Location: US, CA, Santa Clara, US, Remote

Time Type: Full time

Job Description

NVIDIA’s accelerated computing platform is the foundation of modern HPC and AI.At the core of this platform are the CUDA Core Libraries. C++ and Python libraries that enable developers to write fast, reliable, and scalable GPU-accelerated software! We are hiring a full-time Software Engineer to work on the CUDA Core Libraries that power GPU computing for both C++ and Python developers. This includes projects such as CCCL (Thrust, CUB, libcudacxx), cuda-python, and numba-cuda. You will join the team building the foundational libraries, algorithms, and language/runtime infrastructure that make CUDA a speed-of-light experience for developers across deep learning, scientific computing, and data analytics!

What you’ll be doing:

• Develop and implement CUDA Core Libraries in C++ and/or Python, including parallel algorithms and idiomatic language bindings for core CUDA functionality.

• Compose, optimize, and evolve GPU algorithms and APIs, from high-level interfaces down to low-level performance tuning involving memory, parallelism, and synchronization.

• Own features end-to-end: develop, implementation, testing, benchmarking, documentation, and long-term maintenance.

• Improve developer experience across the stack: CI, tests, benchmarks, packaging, examples, and docs.

• Collaborate with senior CUDA engineers in design reviews, code reviews, and open-source-style workflows.

• Engage with real users through issues, performance investigations, and API feedback.

What we need to see:

• BS, MS, or PhD in Computer Science, Computer Engineering, or a related field or equivalent experience.

• Minimum of 8+ years of related development experience

• Strong programming skills in C++, Python, or both, with proven interest in systems-level software (performance, memory, concurrency, API design).

• Solid understanding of modern C++ (templates, generics, standard library) and/or Python library development and packaging.

• Practical experience with parallel or heterogeneous programming (CUDA, OpenMP, GPU-accelerated Python, or similar).

• Experience contributing to production software or open-source libraries, including testing, profiling, and code review.

• Ability to work independently, scope problems, and drive projects to completion.

• Clear written communication for technical design and documentation.

• Comfort navigating large, multi-language codebases (C++, Python, CMake, Pixi, CI systems).

Ways to stand out from the crowd:

• Strong understanding of CPU/GPU architecture and how hardware details affect performance.

• Hands-on experience with CUDA C++, CUDA Python, PyTorch, JAX, Numba, CuPy, or similar GPU-accelerated stacks.

• Familiarity with Thrust, CUB, libcudacxx, or other modern C++/GPU libraries.

• Experience with compiler infrastructure or tooling (LLVM, Clang tooling, MLIR).

• Demonstrated interest in developer tools, library design, and making other developers faster.

If you care deeply about performance, enjoy working at the C++/Python boundary, and want to shape the core CUDA libraries relied on by thousands of developers, this role is a direct fit.

Your base salary will be determined based on your location, experience, and the pay of employees in similar positions. The base salary range is 184,000 USD - 287,500 USD for Level 4, and 224,000 USD - 356,500 USD for Level 5.Applications for this job will be accepted at least until March 19, 2026.

This posting is for an existing vacancy. 

NVIDIA uses AI tools in its recruiting processes.

NVIDIA is committed to fostering a diverse work environment and proud to be an equal opportunity employer. As we highly value diversity in our current and future employees, we do not discriminate (including in our hiring and promotion practices) on the basis of race, religion, color, national origin, gender, gender expression, sexual orientation, age, marital status, veteran status, disability status or any other characteristic protected by law.