Software Engineer - Kernels
MatXJob Title
Software Engineer - Kernels
Role Summary
Design and optimize low-level compute kernels that run on MatX hardware and integrate with the company software stack. Work closely with ML Research and Hardware Engineering to ensure efficient end-to-end execution of large ML workloads.
This role focuses on performance-critical code and close collaboration across teams to translate hardware capabilities into high-performance software primitives.
Experience Level
Mid-level — typically requires 5+ years of software development experience.
Responsibilities
Primary duties include designing, implementing, and tuning kernels and advising on hardware/software integration.
- Design and optimize compute kernels that interface directly with MatX hardware.
- Collaborate with ML Research and Hardware Engineering to define requirements and validate performance.
- Provide programmer-perspective guidance on hardware architecture and software integration.
- Profile, benchmark, and iterate on low-level code for throughput, latency, and correctness.
Requirements
Must-have technical skills, work authorization, and on-site cadence.
- 5+ years of software development experience.
- Proven experience optimizing software for specialized hardware using parallelism, SIMD, assembly-level optimization, or GPU/CUDA techniques.
- Proficiency in at least one low-level language: assembly, C++, C, Zig, or Rust.
- Ability to collaborate with hardware and research teams to validate and tune performance-critical paths.
- Authorized to work in the United States and able to work onsite in Mountain View, CA three days per week (Tuesday–Thursday).
Nice-to-have:
- Experience implementing kernels for ML models (e.g., Transformers).
- Experience with distributed parallelism techniques (AllReduce, AllToAll, data/tensor parallelism).
- Familiarity with compiler internals and code generation for high-performance kernels.
Education Requirements
Bachelor's degree in Computer Science or equivalent degree; equivalent practical experience is accepted.
About the Company
Company: MatX
Headquarters: Mountain View, California, USA
MatX specializes in creating faster chips for large language models (LLMs), focusing on innovative hardware and software solutions. The company fosters a collaborative and supportive work environment, welcoming candidates of all experience levels. Their approach prioritizes deep understanding and consideration of novel methods to drive efficiency and performance in their projects, particularly in silicon design and related engineering roles.
