AI Performance Library Architect

Location: Hybrid work in Hillsboro, Oregon, or Folsom, California, United States; regular on-site presence at an assigned hirify.global site is required.

Annual salary: $170,500–$315,490 USD

Company

hirify.global develops semiconductor technologies and software platforms, including AI-focused products and open-source performance libraries.

What you will do

• Design, develop, and maintain new functionality in the oneDNN open-source library.
• Enable performance-critical portions of AI workloads across hirify.global CPUs and GPUs.
• Support developers optimizing AI frameworks and workloads.
• Collaborate with the cross-platform AI software ecosystem contributing to oneDNN.

Requirements

• Master’s or Ph.D. degree in Mathematics, Physics, Computer Science, or a relevant STEM field.
• 5+ years of experience with C and C++, open-source software, software library architecture, and performance engineering.
• Experience implementing linear algebra algorithms, including BLAS, LAPACK, or PyTorch functions.
• Experience with floating-point arithmetic, numerical stability, Linux development, and low-level optimization using CUDA, x86 assembly or intrinsics, or OpenCL.
• Experienced-hire profile with software development expertise in AI performance optimization.

Nice to have

• Experience with machine learning, deep learning, or high-performance computing applications.
• Experience implementing transcendental functions and non-IEEE low-precision data types such as bfloat16, fp8, or fp4.
• Experience with AI-assisted software development.

Culture & Benefits

• Hybrid work model combining on-site and off-site work.
• Competitive compensation with stock bonuses.
• Health, retirement, and vacation benefits.
• Work on a highly visible open-source project supporting major AI frameworks, including OpenVINO, TensorFlow, PyTorch, and ONNX Runtime.