Senior ML Quantization Engineer for Optical AI Compute

Neurophos, Inc.; Sunnyvale, Northern (TX, KY); Hybrid; USD 150,000 - 240,000; Full time

Benefits offered by this job

Health coverage

HSA contributions

Unlimited PTO

401(k) matching

Stock options

Dental, Vision, Life coverage

Job summary

Neurophos, Inc. seeks an experienced ML engineer to advance post-training quantization for language and diffusion models on our optical inference engines. You will bridge ML research and hardware, enabling customers to deploy AI workloads on Neurophos hardware.

Ideal candidates have deep expertise in model quantization, transformer models, and numerical optimization, with hands-on PyTorch experience and a track record of rigorous experiments. This role is onsite in Austin or Sunnyvale.

Qualifications

• PhD or equivalent research experience in ML, optimization, numerical analysis, or CS.
• 5+ years in ML engineering with at least 3 years focused on model optimization and deployment.
• Experience in neural network quantization, model compression, or efficient inference.
• Strong knowledge of numerical linear algebra, including matrix factorizations and iterative methods.
• Experience with non-convex optimization, discrete optimization, or second-order methods.
• Strong proficiency in PyTorch; familiarity with JAX, Triton, and TensorFlow.
• Hands-on experience with transformer architectures, LLMs, and diffusion models.
• Experience designing controlled numerical experiments and distinguishing algorithmic improvements from artifacts.
• Strong written communication and collaboration skills.

Responsibilities

• Develop hardware-aware post-training methods for full model quantization.
• Investigate preconditioning and formulate quantization as non-convex, discrete, constrained, or second-order optimization.
• Contribute to refining Neurophos's quantization strategy.
• Design controlled numerical experiments to understand potential improvements and secondary effects.
• Build research-quality implementations and reproducible experiment harnesses for testing candidate methods.
• Adapt models from open-source repositories and customer private models.
• Work with models in PyTorch, JAX, Triton, and TensorFlow.
• Design and execute re-quantization, retraining, and other model adaptation techniques.
• Optimize GEMM operations for high-throughput execution.
• Collaborate with hardware, software, and architecture teams to co-optimize model architectures for optical compute characteristics.
• Publish research papers on novel optimization techniques and methodologies, with IP protection.

Skills

Machine learning engineering

Model quantization

Numerical optimization

Transformer models

PyTorch

JAX

Triton

LLMs

Diffusion models

Experimental design

Communication

Education

PhD in ML / related field

Tools

CUDA

NumPy

SciPy

Profiling tools

Job description

Neurophos, Inc. seeks an experienced ML engineer to advance post-training quantization for language and diffusion models on our optical inference engines. You will bridge ML research and hardware, enabling customers to deploy AI workloads on Neurophos hardware.

Ideal candidates have deep expertise in model quantization, transformer models, and numerical optimization, with hands-on PyTorch experience and a track record of rigorous experiments. This role is onsite in Austin or Sunnyvale.