AI Power & Performance Engineer
Core
Evaluate, benchmark, profile, and optimize AI inference workloads across CPU, GPU, APU, and heterogeneous computing platforms.
Role type
Senior IC AI power and performance engineer
Builds
Optimized AI inference performance and power efficiency for AMD processors
Domain
Semiconductor / AI Systems / Heterogeneous Computing
Deliverable
production ML models
Required skills
AI inference workload benchmarking, Linux system-level debugging, performance profiling (perf, Nsight, VTune), Python automation, CPU/GPU architecture knowledge, memory hierarchy optimization, heterogeneous compute utilization
Preferred skills
ROCm, Vitis AI, MIGraphX, model quantization, mixed-precision inference, LLM-serving techniques, MLPerf Inference, Docker, virtualization
Technologies
PyTorch, ONNX Runtime, vLLM, llama.cpp, TensorFlow, ROCm, Nsight Systems, VTune, Perfetto, Linux
Responsibilities
Benchmark and characterize AI inference workloads, Analyze latency, throughput, memory utilization, power consumption, and performance per watt, Profile workloads and identify bottlenecks, Optimize workload performance through quantization and kernel tuning, Conduct competitive benchmarking across platforms, Automate benchmark execution and data collection
Seniority
Senior, hands-on IC
