Senior GPU Networking Architect

NVIDIA seeks a Senior GPU Networking Architect in Switzerland to build the software foundation for global AI systems, linking GPU computing with high-performance networking.

This position sits within the networking software group and is responsible for creating the core communication primitives that bridge GPU hardware capabilities with advanced networking. The architect will help establish the software infrastructure required to support the world’s largest artificial intelligence systems.

Responsibilities

• Design, implement, and refine GPU communication kernels for both collective and point-to-point tasks in large-scale AI environments.
• Utilize a deep understanding of GPU architecture to enhance kernel performance, reduce latency, and ensure computation overlaps with data transfer.
• Create GPU-resident communication primitives and device-side APIs to enable fine-grained, kernel-initiated data movement across nodes and accelerators.
• Profile and tune GPU kernels from start to finish, pinpointing bottlenecks where compute, memory, and network resources intersect.
• Partner with network software, hardware, and AI framework teams to jointly design effective communication strategies.
• Develop proofs-of-concept, run experiments, and perform quantitative modeling to assess new communication approaches.
• Help shape the evolution of programming models that expose GPU-aware networking features to application developers.

Requirements

• At least five years of practical experience with CUDA programming, including writing and optimizing complex GPU kernels.
• An M.Sc. degree or equivalent professional experience in computer science, computer engineering, or a related discipline.
• Solid grasp of GPU architecture fundamentals, including warp scheduling, shared memory, L2 cache, memory coalescing, occupancy tuning, and asynchronous execution.
• Proven background in systems-level C/C++ development within performance-sensitive settings.
• Familiarity with GPU data movement mechanisms like GPUDirect RDMA and GPU-initiated communication.
• Ability to analyze GPU performance profiles using tools such as Nsight Compute or Nsight Systems and convert findings into concrete optimizations.
• Strong collaborative skills capable of thriving in a multi-national, interdisciplinary environment.

Nice to have

• Prior experience building or optimizing communication kernels in libraries like NCCL, NVSHMEM, or comparable GPU-aware communication frameworks.
• Knowledge of distributed deep learning parallelism methods, including data, tensor, pipeline, expert, and mixture-of-experts parallelism.
• Background in RDMA, InfiniBand, high-speed networking, and GPU system topologies such as NVLink, NVSwitch, PCIe, and network fabrics.
• Experience with latency-hiding techniques like kernel pipelining, persistent kernels, or cooperative groups.
• Demonstrated history of evaluating and optimizing large-scale LLM training or inference workloads, including hands-on use of PyTorch, TensorRT-LLM, or vLLM.

What the company offers

• Highly competitive salary packages.
• A comprehensive benefits package.
• A diverse and supportive work environment that encourages employees to perform at their best.

About the company

NVIDIA is recognized as one of the most desirable employers in the technology sector, driven by a unique legacy of innovation fueled by exceptional technology and talented individuals. Employees collaborate with colleagues who bring deep expertise and innovative thinking to the industry.

• Work alongside experts who demonstrate deep industry knowledge and innovative thinking.
• Join a company widely regarded as one of the most desirable employers in the technology world.
• Benefit from a legacy of innovation supported by advanced technology and talented staff.