Advanced AI Workloads - Performance and Scalability Engineer
Core
Principal Engineer leading a team to drive performance and scalability of frontier AI workloads (LLMs, MoE, diffusion models) on AMD GPUs across the full stack from kernels to distributed systems.
Role type
Principal, hands-on technical team lead
Builds
High-performance AI inference and training systems on AMD GPU hardware
Domain
Semiconductor / AI Systems / High-Performance Computing
Deliverable
production ML models
Required skills
GPU micro-architecture, C++ (C++17+), Python, kernel optimization, compiler toolchains (LLVM, MLIR), distributed systems, performance profiling, open-source framework contribution
Preferred skills
PhD in CS/EE, experience with HIP/CUDA/OpenCL/Triton, multi-GPU communication topologies, Linux systems, published research
Technologies
AMD GPUs, vLLM, SGLang, PyTorch, CUDA, HIP, Triton, CUTLASS, CK, LLVM, MLIR, RDMA, NCCL, RCCL, Infinity Fabric, xGMI, NVLink
Responsibilities
Lead a small team of engineers, define long-term technical strategy, own complex cross-stack performance challenges, design novel GPU kernels and compiler optimizations, establish performance methodology, influence upstream open-source AI frameworks, drive architecture decisions for emerging inference paradigms, identify performance gaps vs competitors, mentor engineers, represent AMD externally
Seniority
Principal, hands-on IC with team leadership
