← All roles

Senior GPU Inference Performance Engineer

AMD· Engineering
Apply on company site
Santa Clara, CA$164kFull-timePosted 3mo ago

About the role

Senior-level role focusing on end-to-end performance analysis of GPU-accelerated AI inference workloads, profiling across hardware, software runtimes, and networking to drive competitive performance improvements.

Responsibilities

  • Full-stack GPU profiling: Instrument and analyze inference workloads across AMD Instinct (ROCm, rocProfiler, ROCm Systems Profiler, RGP, rocprof-compute, rocprof-sys, Omniperf) and NVIDIA (CUDA, Nsight Systems/Compute, DCGM) GPUs. Identify bottlenecks spanning HBM bandwidth, compute utilization, kernel scheduling, memory allocation, PCIe/Infinity Fabric data movement, and GPU runtime behavior.
  • Systems and runtime performance analysis: Profile and diagnose interactions between GPU runtimes, Linux OS, device drivers, container runtimes, memory subsystems, CPU scheduling, NUMA topology, and I/O pathways.
  • Competitive performance analysis: Design and execute head-to-head benchmarks (AMD vs. NVIDIA) on standardized AI and LLM workloads; provide data-backed explanations for performance differences attributing to hardware architecture, networking topology, libraries, or configurations.
  • Multi-server inference networking: Profile and optimize distributed inference topologies across multi-node clusters; analyze RDMA/RoCE traces, NCCL/RCCL profiling, GPUDirect RDMA, NIC counters, and network performance tools; quantify impact on end-to-end inference SLAs.
  • GPU operator and Kubernetes stack: Profile overhead from GPU operators, device plugins, container runtimes (Docker, containerd), and Kubernetes scheduling; identify jitter, cold-start, resource contention, and infrastructure inefficiencies in production.
  • Tooling and automation: Build reproducible benchmarking harnesses, profiling scripts, performance regression dashboards; automate trace collection and analysis for firmware, drivers, networking stacks, runtimes, and AI frameworks.

Qualifications

  • Background in GPU performance engineering, HPC, distributed systems, networking, OS, or systems performance analysis.
  • Hands-on proficiency with AMD (ROCm, rocProfiler, ROCm Systems Profiler, RGP, rocprof-compute, rocprof-sys, Omniperf/Omnitrace) or NVIDIA (CUDA, Nsight Systems/Compute, NCU) profiling toolchains; deep understanding of GPU architecture, including warp/wavefront execution, memory hierarchy, occupancy, and instruction-level parallelism.
  • Experience analyzing GPU communication and networking performance including NCCL/RCCL, RDMA/RoCE, GPUDirect RDMA, UCX, MPI, ConnectX, Pensando, or similar high-performance networking technologies.
  • Experience with multi-GPU and multi-node inference, training, or HPC environments including tensor parallelism, pipeline parallelism, distributed communication libraries, and network performance analysis tools.
  • Experience with Linux systems performance analysis, operating systems, device drivers, virtualization, container runtimes, or low-level systems software development.
  • Demonstrated ability to explain performance differences in written reports or presentations rooted in hardware and software evidence.
  • Strong Python and C/C++ skills; comfortable reading GPU kernel code (HIP/CUDA), runtime code, or systems-level software.
  • Experience with Kubernetes GPU scheduling, MIG, GPU operator performance, or contributions to open-source infrastructure, systems, networking, inference, or profiling projects.

Skills

Required:GPU performance profilingLinux systems performance analysisPythonC/C++
Nice to have:NCCL/RCCL, RDMA/RoCE, GPUDirect RDMAKubernetes GPU scheduling, MIG
Last updated Oct 15, 2026. Interested? Apply on the company site.
AMD logo

About AMD

together we advance_AI

AMD delivers leadership high-performance and adaptive computing solutions to advance data center AI, AI PCs, intelligent edge devices, gaming, & beyond.

AIData CenterHigh-Performance ComputingGamingEdge Devices
Company profileWebsite

More open roles at AMD