Senior/Staff AI Model Optimization Architect (Hsinchu/Taipei)
Core
Lead end-to-end model transformation and optimization for LLMs, VLMs, diffusion, and multimodal models on Qualcomm inference accelerators.
Role type
Staff AI Model Optimization Architect
Builds
Efficient inference execution for large-scale foundation models on Qualcomm hardware
Domain
Cloud AI, Machine Learning, Computer Architecture
Deliverable
production ML models
Required skills
PyTorch, graph capture and compilation (torch.compile/TorchDynamo), transformer architectures, KVcache management, distributed inference, fusion kernels (Triton), Python engineering
Preferred skills
Triton kernel development, LLM serving stacks, continuous batching, numerical methods analysis, PhD in relevant field
Technologies
PyTorch, ONNX, Triton, torch.compile
Responsibilities
Architect model optimization strategies, drive graph capture and deployment, design fusion kernels, partner with compiler teams, profile and optimize inference, own transformer optimizations, enable continuous batching, architect distributed inference strategies, debug performance issues
Seniority
Staff, hands-on IC
