Principal Software Engineer Kernels
d-MatrixVisit d-Matrix website
AI-infrastructure company producing memory-centric hardware, networking, and software for low-latency generative-AI inference in data centers.
Santa Clara, United States
Funding history
About d-Matrix
d-Matrix develops digital in-memory-compute technology and a full-stack inference platform comprising Corsair accelerators, JetStream networking accelerators, and Aviator software.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will develop, enhance, and maintain kernels for AI hardware. You will map algorithms and computational graphs to hardware architectures, optimize hardware-software tradeoffs, and work with compiler, ML, systems, and hardware specialists on the software toolchain.
Requirements
- MS in computer engineering, math, physics, or a related field with 10+ years of industry experience, or a PhD with 5+ years of industry experience
- Knowledge of computer architecture, data structures, system software, and machine learning fundamentals
- Proficiency in C, C++, and Python development in Linux environments
- Experience implementing algorithms in C, C++, and Python
- Experience implementing algorithms for FPGAs, DSPs, GPUs, and AI accelerators using CUDA
- Experience implementing ML operators including GEMMs, convolutions, BLAS, SIMD operations, softmax, layer normalization, and pooling
- Experience developing for embedded SIMD vector processors such as Tensilica
Responsibilities
- Develop, enhance, and maintain software kernels for next-generation AI hardware
- Map algorithms to hardware architectures
- Map AI-framework computational graphs to underlying architectures
- Optimize hardware-software co-design tradeoffs
- Build and scale software deliverables
- Collaborate on compiler infrastructure with compiler, ML, systems, and hardware specialists
Benefits
- Equity
- Bonus
- Medical insurance
- Dental insurance
- Vision insurance
- 401(k)