AI Compiler R&D Expert

Forestown; California City (CA); On-site; USD 100,000 - 130,000; Full time

Job summary

A leading tech company in California seeks a highly skilled individual to implement and optimize compilers for vector-related components in AI chips. The ideal candidate will have a Master’s degree in Computer Science, over three years of experience, and proficiency in C/C++. Familiarity with heterogeneous programming languages like CUDA and experience in leading a team are essential. Join us to push the boundaries of AI technology.

Qualifications

• More than three years of relevant work experience in AI development.
• Strong understanding of compiler principles and architectures.
• Experience leading a team of about 10 people.

Responsibilities

• Implement and optimize the compiler for vector-related components.
• Enhance the AI heterogeneous computing framework.

Skills

C/C++ programming

Compiler development

Heterogeneous programming (CUDA/OpenCL/Triton)

Education

Master's degree in Computer Science or related major

Tools

LLVM

PyTorch

Job description

Responsibilities

1. Implement and optimize the compiler for vector-related components in AI chips.
2. Implement and optimize the AI heterogeneous computing framework.

Qualifications

• Master's degree or above in Computer Science or a related major, with more than three years of relevant work experience.
• Familiar with computer architecture and related quantitative analysis methods, compiler principles, and compilation frameworks such as LLVM.
• Proficient in C/C++ programming language, with rich experience in compiler development, optimization, and debugging.
• Familiar with heterogeneous programming languages such as CUDA/OpenCL/Triton and their compilation implementations.
• Familiar with PyTorch, OpenXLA, etc.
• Have relevant R & D experience in heterogeneous programming environments.
• Work experience in chip compiler/toolchain technology R & D and architecture design at Huawei/NVIDIA is preferred.
• Capable of leading a team of about 10 people.