Compiler Frontend Engineer
Core
Extend the compiler frontend and IR to support new LLM architectures (attention variants, MoE, quantization) for in-memory inference hardware.
Role type
Senior IC compiler frontend engineer (ML)
Builds
Compiler passes, IR extensions, and lowering logic for novel LLM architectures
Domain
AI infrastructure / Semiconductor / ML compilers
Deliverable
production ML models
Required skills
ML compiler development, PyTorch internals (TorchDynamo, FX, torch.export), LLM architecture knowledge (attention, MoE, quantization), Python, C++ or Rust
Preferred skills
IREE, XLA, TVM, JAX, ONNX, tensor/pipeline/expert parallelism, hardware-software co-design
Responsibilities
Implement frontend support for new LLM architectures, extend IR and lowering passes, validate models against simulators and FPGA prototypes, track LLM architecture landscape, hand off lowered programs to backend engineers
Seniority
Senior, hands-on IC
