Compiler Frontend Engineer

Job Title

Compiler Frontend Engineer

Role Summary

The Compiler Frontend Engineer will extend the compiler IR, ops, and lowering passes to support new LLM architectures and bring models up on Normal's pre-silicon stack. This role works closely with backend compiler engineers, LLM serving engineers, and hardware teams to ensure new architectures run correctly on simulators, FPGA prototypes, and, later, silicon.

Research-adjacent position: you will follow literature closely and publish on behalf of the company. Accessibility accommodations: contact accommodations@normalcomputing.com. See the Applicant Privacy Notice.

Experience Level

Mid-level. The posting does not state explicit years, but expects prior experience shipping ML compiler passes and performing model bring-up.

Responsibilities

Primary responsibilities focus on representing new LLM primitives in the frontend and ensuring correctness through validation and handoff to backends.

• Bring up frontend and mid-level compiler support for new LLM architectures (attention variants, MoE routing, etc.).
• Extend IR, ops, and lowering passes so architectural primitives are represented faithfully with test coverage.
• Model quantization and numeric-format representation to preserve model accuracy through lowering.
• Track LLM architecture developments and recommend priorities for bring-up.
• Validate compiler behavior on simulators and FPGA prototypes and hand lowered programs to backend teams for hardware mapping.
• Collaborate with serving engineers to transition supported models into production-serving stacks.

Requirements

Must-have technical skills and capabilities for successful performance in this role.

• Hands-on ML compiler experience: written and shipped compiler passes.
• Deep familiarity with PyTorch internals; ideally experience with TorchDynamo, FX, or torch.export.
• Working knowledge of modern LLM architectures, including attention mechanisms, MoE routing, and quantization schemes.
• Fluency in Python; familiarity with C++ or Rust.
• Experience building software for hardware in development, validating on simulators and FPGA prototypes.
• Ability to design and extend compiler infrastructure in an evolving codebase and to follow research literature closely.

Nice-to-have:

• Experience with other ML compilers (IREE, XLA, TVM) or graph formats (JAX, ONNX).
• Experience partitioning models across devices (tensor/pipeline/expert parallelism) or hardware-software co-design.
• Experience targeting non-standard or in-memory hardware and contributions to open-source compiler/ML projects.

Education Requirements

Not specified.

About the Company

Company: Normal Computing

Headquarters: New York, NY, USA

Normal Computing develops software and hardware solutions for the semiconductor and AI infrastructure industries, specializing in ASICs for image and video diffusion inference and AI accelerators. The company focuses on architecture and microarchitecture of compute blocks, PE array design, ISA co-design, and FPGA prototyping.