Microsoft is looking to pioneer kernel fusion compilation techniques, develop advanced AI graph compilation tools, and design new programming models for its next-generation supercomputers. A key focus is the automatic fusion and tuning of Triton kernels.
Requirements
- coding in languages including, but not limited to, C, C++, C, Java, JavaScript, or Python
- 2+ years experience with C/C++
- Experience building compilers, using compiler frameworks like LLVM/MLIR, or optimizing AI/numerical workloads
- Experience with graph compilation or kernel fusion
- 2+ years experience with Python
- Experience using or developing Machine Learning training or inference software
- A strong technical background and solid foundation in software engineering principles.
Responsibilities
- Design and develop AI software in C/C++, Python, and other languages.
- Designing and building polyhedral analysis and graph compilation tools for large-scale AI model fusion and optimization
- Implementing innovative new graph and fusion compiler features and optimization passes
- Developing code generation techniques for novel hardware platforms
- Designing new programming abstractions for AI
- Collaborating broadly across multiple disciplines from hardware architects to ML developers.
- Help establish and drive the adoption of outstanding coding standards and patterns and help enhance our inclusive engineering culture.
Other
- A strong commitment to a diverse and inclusive workgroup culture
- Strong intellectual curiosity and passion about learning new technologies.
- Great cross-team collaboration skills and communication skills
- Identify requirements, scope solutions, estimate work, schedule deliverables.
