Senior Architect - Server Performance

We are seeking architects to drive architectural performance for our next-generation AI server systems. This position demands a unique capability to bridge deep architectural knowledge, workload analysis, and hands-on silicon investigations.

Responsibilities include conducting performance investigations on both NVIDIA and competitive platforms, and developing targeted microbenchmarks to examine specific architectural aspects.

As a Senior Architect - Server Performance, you will:

• Analyze workloads of interest on existing silicon, with an emphasis on at-scale AI workloads, and high-performance computing (HPC) applications.
• Collaborate with cross-functional teams to define performance metrics and key use-case scenarios, then develop robust tests and benchmarking methodologies.
• Conduct comprehensive performance evaluations, identify bottlenecks, and recommend effective solutions using appropriate tools and platforms.
• Utilize insights from workload analysis and silicon studies to propose architectural features that optimize system performance and scalability.
• Work closely with software and hardware teams to influence design choices that impact overall system performance.
• Act as a subject matter expert on system performance, providing guidance and support to the broader engineering team.

Requirements include:

• Bachelor’s or Master’s degree in a relevant field; a PhD is a plus.
• 10+ years of practical experience in hardware architecture across areas such as CPU, GPU, cache, memory subsystem, PCIe, networking, or storage.
• Expertise in high-performance networking technologies, including InfiniBand and RoCE, or a strong familiarity with communication libraries like MPI and UCX.
• Alternatively, a proven track record in performance optimizations for deep learning training or inference systems, high-performance computing, or cloud computing environments.
• Expertise in benchmarking tools and methodologies, with demonstrated skill in developing and implementing targeted microbenchmarks.
• Solid understanding of performance analysis tools and techniques; experience with performance simulators is highly desirable.
• Proficiency in programming languages such as C, C++, and Python.
• Strong ability to work within large, complex, and unfamiliar software repositories.