Senior ML Engineer — Optical AI Inference & Quantization

Socket.dev; Austin (TX); On-site; USD 150,000 - 230,000; Full time

Benefits offered by this job

Health premium coverage

Unlimited PTO

401(k) matching

Stock options

Voluntary benefits

Job summary

Neurophos in Austin, TX or Sunnyvale, CA is seeking an experienced machine learning engineer to develop advanced post-training quantization methods for large language models, diffusion models, and ML workloads on our metamaterial optical inference engines.

This role bridges cutting-edge ML research with novel hardware, ensuring customers can deploy AI workloads on Neurophos hardware, with opportunities to publish research and collaborate across hardware, software, and architecture teams.

Qualifications

• PhD or equivalent research experience in ML/optimization or related field.
• 5+ years in ML engineering with 3+ years in model optimization and deployment.
• Experience in neural network quantization, model compression, or efficient inference.
• Strong knowledge of numerical linear algebra and iterative methods.
• Experience with non-convex or discrete optimization methods.
• Proficiency in PyTorch and familiarity with JAX, Triton, TF.

Responsibilities

• Develop hardware-aware post-training methods for full model quantization.
• Investigate preconditioning and quantization optimization techniques.
• Refine Neurophos's quantization strategy and experiments.
• Design controlled numerical experiments to evaluate improvements.
• Build research-grade implementations and reproducible experiment harnesses.
• Adapt models from open-source and customer private models.
• Work with PyTorch, Triton, JAX, and new frameworks.
• Perform re-quantization, retraining, and precision-reduction techniques.
• Optimize GEMM operations for high throughput.
• Collaborate with hardware, software, and architecture teams.
• Publish research papers on optimization methods.

Skills

Model optimization

Post-training quantization

Hardware-aware ML

Quantization strategies

Experiment design

PyTorch

JAX

Triton

Transformer models

LLMs

Education

PhD or equivalent research experience

Tools

PyTorch

JAX

Triton

TensorFlow

Job description

Neurophos in Austin, TX or Sunnyvale, CA is seeking an experienced machine learning engineer to develop advanced post-training quantization methods for large language models, diffusion models, and ML workloads on our metamaterial optical inference engines.

This role bridges cutting-edge ML research with novel hardware, ensuring customers can deploy AI workloads on Neurophos hardware, with opportunities to publish research and collaborate across hardware, software, and architecture teams.