Staff Compiler Engineer

Job Title

Staff Compiler Engineer

Role Summary

Lead development of the compiler toolchain for Neurophos's custom LLM accelerator, implementing compiler components and optimizations that map ML workloads to hybrid optical‑electronic hardware.

Work with compiler backends, ML framework integrations, and hardware teams to deliver correct, high-performance inference on the target architecture. Onsite role in Austin or the Bay Area.

Experience Level

Senior (Staff level). Requires 5+ years in compiler development; 7+ years preferred.

Responsibilities

Design and build compiler components and optimization passes that map ML models to custom accelerator hardware, and maintain correctness and performance validation.

• Design and implement toolchains for a custom LLM accelerator architecture.
• Develop optimization strategies bridging high-level ML algorithms to hardware implementations.
• Design and implement IR dialects, graph transformations, and lowering passes.
• Optimize computational graphs and memory access patterns for the target architecture.
• Integrate compiler toolchain with ML frameworks (e.g., PyTorch, JAX, Triton).
• Build and maintain test infrastructure and performance regression suites to ensure correctness and performance.
• Collaborate with hardware, systems, and ML teams on co‑design and verification.

Requirements

Core technical skills and hands-on compiler and systems experience required; preferred skills listed separately.

• 5+ years of compiler development experience, including backend/optimization work.
• Expert-level proficiency in Python and C.
• Experience with hardware compilers and compiler frameworks (LLVM, GCC, custom backends); MLIR experience is a plus.
• Familiarity with Large Language Model architectures and their computational requirements.
• Strong understanding of computer architecture, memory hierarchies, and parallel computing concepts.
• Experience with AI/ML accelerators (GPUs, TPUs, FPGAs) and relevant programming models.
• Practical experience with testing, CI/CD, and performance benchmarking.
• Development tooling: CMake, Git, Docker; experience with profilers and parallel debugging tools.

Nice-to-have:

• Experience with graph algorithms, graph rewriting systems, and DAG optimization.
• Experience parsing, analyzing, and mutating programs (AST processing).
• Experience with quantization techniques, model optimization, and low-latency/HPC system design.
• Experience instrumenting and debugging parallel programs and production inference engines.
• Familiarity with structured, human-supervised AI coding workflows.

Education Requirements

Master's degree in Computer Science or a related technical field is required. PhD in Computer Science or related field is preferred.

About the Company

Company: Neurophos

Headquarters: Sunnyvale, California, United States

Startup developing silicon-photonic AI accelerator chips that use programmable metasurfaces and dense optical cells to perform matrix multiplications at the speed of light, targeting large-scale AI inference with much higher energy efficiency and performance than traditional electronic approaches.