Software Engineer, Staff - SIMD Kernels
Core
Develop, enhance, and maintain software kernels for ML operators (softmax, layer norm, activation functions) for next-generation AI hardware and build intuitive SDKs for developers.
Role type
Staff Software Engineer (SIMD Kernels & AI Compute)
Builds
Software stack for AI compute engine, ML operator kernels, and developer SDKs
Domain
AI hardware, computer architecture, machine learning systems
Deliverable
production ML models
Required skills
C/C++, Python, Linux development, computer architecture, data structures, ML fundamentals, algorithm implementation for specialized hardware (FPGAs, DSPs, GPUs, AI accelerators), CUDA, ML operators (GEMMs, Convolutions, pooling)
Preferred skills
Startup/incubation experience, ML frameworks (TensorFlow, PyTorch), ML compilers (MLIR, LLVM, TVM, Glow), deep learning models (CV, NLP, recommendation), embedded SIMD vector processors (Tensilica), cloud provider or AI compute company experience
Technologies
CUDA, TensorFlow, PyTorch, MLIR, LLVM, TVM, Glow, Tensilica, Linux
Responsibilities
Develop and maintain software kernels for ML operators, build solutions to make SDK intuitive for developers, analyze performance, map algorithms and AI-framework computational graphs onto hardware architectures, navigate hardware–software co-design trade-offs
Seniority
Staff, hands-on IC with leadership
