英伟达Senior GPU Architect

• Investigate and design new hardware features for future graphics and parallel processing architectures. • Work in a team to document, design, develop tools to analyze and simulate, validate, and verify functional or performance models. • Develop tests, testplans, and testing infrastructure for new graphics or parallel processing architectures • Be hungry to learn and work on simulators, RTL and real silicon.

• Lead presales and architecture engagements with AI industry customers, focusing on GPU servers, AI clusters, and large‑scale training/inference platforms built on NVIDIA HGX, GPU systems, and reference architectures. • Design and validate end‑to‑end AI data center solutions, including server platforms, storage connectivity, and high‑performance networking based on Spectrum, Quantum, ConnectX, and BlueField. • Define system architectures for AI supercomputing, LLM training, and inference workloads, including node configuration, GPU topology, PCIe/NVLink considerations, and network design. • Support business teams in exploring, developing, and deploying NVIDIA server and GPU solution opportunities, from early technical discovery through POC and production rollout. • Own and execute POCs and hands‑on labs that validate GPU server performance, scalability, reliability, and interoperability across compute, storage, and network domains. • Troubleshoot complex end‑to‑end issues involving GPU servers, firmware, drivers, operating systems, and networking stacks, and drive fixes with internal R&D and partners. • Provide structured feedback on platform features, system requirements, and customer needs to server OEMs, engineering, and product teams to improve NVIDIA AI platforms and ecosystems.

ResponsibilitiesDesign, develop, and optimize high-performance AI operators and GPU kernels for training and inference workloads.Analyze and improve kernel performance through profiling, memory optimization, kernel fusion, parallelization, and architecture-aware tuning.Develop GPU kernels using CUDA, HIP, Triton, or similar programming frameworks.Build AI-powered coding agents and automation workflows for kernel generation, optimization, benchmarking, and validation.Collaborate with model and infrastructure teams to improve end-to-end AI system performance.RequirementsBachelor's degree or above in Computer Science, Computer Engineering, Artificial Intelligence, or a related field, with 4+ years of relevant experience.Strong programming skills in C++ and/or Python.Hands-on experience in GPU kernel development and optimization using CUDA, HIP, Triton, or similar frameworks.Solid understanding of GPU architecture and performance optimization, including memory hierarchy, occupancy, register pressure, and shared memory.Experience with performance profiling and optimization tools such as Nsight, rocprof, or equivalent.Experience with AI operators such as GEMM, Attention, MoE, Embedding, or other performance-critical workloads.Preferred QualificationsDeep experience in GPU kernel optimization, including kernel fusion, memory optimization, communication/computation overlap, performance modeling, or optimization of AI operators such as GEMM, Attention, MoE, and Embedding.Experience building coding agents, code generation systems, or AI-assisted software engineering tools.Experience with large-scale AI training or inference systems.Nice-to-Have Experience with LLM post-training, including SFT, RLHF, DPO, GRPO, or related techniques.Publications or open-source contributions in GPU optimization, AI systems, or coding agents. #LI-JW2

NVIDIA has been transforming computer graphics, PC gaming, and accelerated computing for more than 25 years. It’s a unique legacy of innovation that’s fueled by great technology—and amazing people. Today, we’re tapping into the unlimited potential of AI to define the next era of computing. An era in which our GPU acts as the brains of computers, robots, and self-driving cars that can understand the world. Doing what’s never been done before takes vision, innovation, and the world’s best talent. As an NVIDIAN, you’ll be immersed in a diverse, supportive environment where everyone is inspired to do their best work. Come join the team and see how you can make a lasting impact on the world.  What you’ll be doing: • Be responsible for running test cases to validate NVIDIA GPU Communications Libraries (NCCL, NVSHMEM, UCX, GDRCopy, GPUDirect RDMA etc). • Be responsible to automate test cases and maintain the automation scripts. • Collaborate with Developer, PM, marketing, and engineering teams on crafting test plan and implementing validation. • You will assist in the architecture, crafting and implementing of SWQA test frameworks. • Be responsible for code coverage improvement and code complexity optimization.