Job Summary:
Neurophos is an innovative startup redefining the future of AI hardware through silicon photonics technology. They are seeking a talented ML Compiler Engineer to lead the development of their compiler for novel LLM accelerator architecture, bridging high-level AI workloads with custom hardware.
Responsibilities:
โข Design and implement toolchains for our custom LLM accelerator architecture
โข Develop optimization strategies that bridge software algorithms to hardware implementations
โข Design and implement custom compiler components, including IR dialects, graph transformations, and lowering passes
โข Optimize computational graphs and memory access patterns for our hardware architecture
โข Integrate with existing ML frameworks (e.g., PyTorch, JAX, Triton)
โข Build and maintain test infrastructure to ensure compiler correctness and performance
Qualifications:
Required:
โข Bachelor's degree in Computer Science, Computer Engineering, or related field
โข 10+ years of industry experience
โข 5+ years of professional experience in systems programming or compiler development
โข Expert-level proficiency in Python and C
โข Experience with hardware compilers
โข Familiarity with Large Language Model architectures and their computational requirements
โข Hands-on experience with compiler frameworks and code optimization techniques
โข Deep understanding of computer architecture, memory hierarchies, and parallel computing concepts
โข Experience with AI/ML accelerators (GPUs, TPUs, FPGAs) and their programming models
Preferred:
โข Master's degree in Computer Science, Computer Engineering, or related field
โข Strong background in graph theory and graph transformations in a compiler or optimization context; MLIR experience is a plus
โข Experience writing programs that parse, analyze, and mutate programs as abstract syntax trees
โข Experience in instrumenting and debugging parallel programs
โข Experience with structured, human-supervised AI/agentic coding workflows
โข Experience with LLM quantization techniques and model optimization
โข Experience with high-performance computing and low-latency system design
โข Familiarity with deep learning frameworks and neural network optimization
Company:
Neurophos develops photonic AI processing technology that focuses on hardware solutions for accelerating artificial intelligence inference. Founded in 2020, the company is headquartered in Austin, USA, with a team of 11-50 employees. The company is currently Early Stage.