ML Optimization Engineer: Training & Inference on GPUs

Generalist; Somerville (NJ); On-site; USD 150,000 - 210,000; Full time

Job summary

Generalist in New Jersey is seeking a highly skilled ML systems engineer to push the boundaries of training efficiency and real-time robot inference. You will own optimizations across GPUs, software stacks, and hardware interfaces to unlock step-function performance gains.

You will collaborate with researchers and engineers to prototype, benchmark, and deploy models at scale, focusing on transformer and diffusion architectures while minimizing latency and power usage.

Qualifications

• Proficient in ML training and inference optimizations for transformer and diffusion models.
• Experience optimizing GPU and system level bottlenecks.
• Strong coding in Python; familiarity with CUDA is a plus.

Responsibilities

• Improve both training and on-robot inference stacks.
• Implement ML, hardware, and software changes for step-function gains.
• Tune CUDA kernels and data pipelines to remove bottlenecks.

Skills

ML optimization

Transformer models

Diffusion models

CUDA kernels

Python programming

Tools

PyTorch

NumPy

CUDA

Job description

Generalist in New Jersey is seeking a highly skilled ML systems engineer to push the boundaries of training efficiency and real-time robot inference. You will own optimizations across GPUs, software stacks, and hardware interfaces to unlock step-function performance gains.

You will collaborate with researchers and engineers to prototype, benchmark, and deploy models at scale, focusing on transformer and diffusion architectures while minimizing latency and power usage.