Senior GPU Compiler Development Engineer
Core
Architect and implement MLIR compiler infrastructure to lower deep-learning operators for next-generation large language and vision models onto AMD GPUs.
Role type
Senior IC MLIR compiler engineer (GPU)
Builds
MLIR dialects, operator-lowering surfaces, and a new single dynamic-shape graph for AI workloads
Domain
AI/ML compilers, GPU architecture, intermediate representations
Deliverable
production ML models
Required skills
C/C++, MLIR (dialects, lowering, passes), deep-learning operators, computational graphs, LLVM, GPU programming (HIP/CUDA), ONNX/ONNX Runtime, dynamic-shape inference, quantized inference
Preferred skills
Attention/transformer/diffusion operator internals, Windows/Linux development, debuggers, profilers, AI coding agents
Technologies
MLIR, LLVM, HIP, CUDA, ONNX, GitHub, Linux, Windows
Responsibilities
Architect dynamic-graph IR lowering, implement operator definitions and MLIR lowerings for attention/control-flow, build runtime-status reporting, contribute to MLIR dialect conventions, debug and research optimization strategies
Seniority
Senior, hands-on IC
