Frontend ML Compiler Engineer for LLMs & Hardware

Digitailor; Palo Alto (CA); On-site; USD 140,000 - 230,000; Full time

Job summary

Normal Computing in Silicon Valley builds AI hardware/software for the semiconductor industry. We are seeking a Compiler Engineer focused on frontend and model bring-up to extend our IR, ops, and lowering passes for new LLM architectures.

You will validate bring-up against simulators and FPGA prototypes, ensuring models run correctly ahead of silicon arrivals, and collaborate with backend teams to map models onto our hardware.

Qualifications

• Hands-on ML compiler experience and shipping compiler passes.
• Deep familiarity with PyTorch internals and compilation stack (TorchDynamo, FX, torch.export).
• Knowledge of modern LLM architectures: attention, MoE routing, quantization.

Responsibilities

• Model Bring-up: extend frontend/mid-level support for new LLM architectures in our compiler.
• IR and Lowering: extend IR/ops/lowering passes with test coverage.
• Quantization and Numerics: represent new quantization schemes for accuracy.
• Architecture Tracking: monitor LLM landscape to plan next bring-ups.
• Cross-Team Handoffs: coordinate with backend compiler engineers and LLM serving.

Skills

ML compiler

PyTorch internals

LLM architectures

Quantization schemes

Attention mechanisms

Rust/C++ familiarity

Tools

TorchDynamo

FX

torch.export

Job description

Normal Computing in Silicon Valley builds AI hardware/software for the semiconductor industry. We are seeking a Compiler Engineer focused on frontend and model bring-up to extend our IR, ops, and lowering passes for new LLM architectures.

You will validate bring-up against simulators and FPGA prototypes, ensuring models run correctly ahead of silicon arrivals, and collaborate with backend teams to map models onto our hardware.