Back to all jobs
Etched
Inference Software Engineer
Cupertino CA
About this role
Etched is seeking a Inference Software Engineer to join their AI Compute & Systems Software team. This role involves cutting-edge work in AI acceleration and semiconductor technologies. The ideal candidate will have deep expertise in their area of specialization and a passion for pushing the boundaries of hardware and software co-design.
What you’ll do
- Build and optimize kernels, compilers, runtimes or firmware for AI accelerators
- Profile and remove performance bottlenecks across the hardware/software stack
- Integrate with ML frameworks such as PyTorch, JAX or Triton
Required silicon / compute stack
- C/C++
- CUDA
- LLVM / MLIR
- Triton
- PyTorch
- Kernels & Runtimes
Requirements
- Strong C/C++ and systems programming skills
- Experience with CUDA, accelerator programming, compilers (LLVM/MLIR) or low-level drivers
- Understanding of computer architecture and parallel performance optimization
Typical for this role category. See the employer’s official posting for the exact requirements.
About Etched
Etched is building Sohu, a transformer-specific ASIC designed to deliver far higher throughput for transformer inference than general-purpose GPUs.
Apply Now
You’ll apply on the employer’s official site, which opens in a new tab. ChipTalent.io never collects your application.