Fellow GPU Performance Optimization Engineer
Core
Maximize performance and efficiency of large-scale AI training workloads on AMD GPU platforms by driving innovations across the full software-hardware stack.
Role type
Fellow GPU Performance Optimization Engineer (AI Training Systems)
Builds
High-throughput, scalable generative AI training systems on AMD hardware
Domain
AI/ML Infrastructure, High-Performance Computing, GPU Systems
Deliverable
production ML models
Required skills
GPU architecture expertise, distributed training optimization, system-level bottleneck analysis, kernel and runtime optimization, performance profiling, C++/CUDA/HIP programming, ML framework tuning (PyTorch/JAX/TensorFlow), communication library optimization (NCCL/RCCL)
Preferred skills
Compiler stack optimization, graph-level optimization, experience with Megatron-LM/Torchtitan/MaxText, technical leadership
Technologies
ROCm, Nsight, PyTorch, JAX, TensorFlow, NCCL, RCCL, C++, CUDA, HIP, Python
Responsibilities
Lead performance optimization of large-scale AI training workloads across single-node and multi-node environments; Identify and eliminate system bottlenecks across compute, memory, and communication; Optimize distributed training strategies (Data, Tensor, Pipeline Parallelism, ZeRO); Drive cross-stack optimizations spanning kernels, compilers, runtimes, and frameworks; Develop advanced profiling and benchmarking methodologies; Collaborate with hardware and compiler teams to influence architecture design; Lead open-source efforts to improve ecosystem performance
Seniority
Fellow, strategic technical leadership & hands-on IC
