ML Systems Engineer: Optimizing Training & GPU Kernels

Jobtailor; Massachusetts; On-site; USD 120,000 - 180,000; Full time

Job summary

Jobtailor in Massachusetts seeks an experienced ML engineer to optimize training and inference across large models. You will profile data loading and kernel performance, implement fusion, tiling, and sharding, and collaborate with distributed training teams.

Candidates should excel in Python and PyTorch, have deep ML fundamentals, and demonstrate strong analytical problem-solving with a bias for action. Experience with CUDA or Triton is preferred, with a track record of throughput improvements.

Qualifications

• Advanced knowledge of Python and PyTorch for ML workloads.
• Experience optimizing training and inference execution.
• Strong problem-solving and data-driven approach.

Responsibilities

• Profile and optimize data loading to maximize throughput.
• Develop high-performance GPU kernels in CUDA/Triton.
• Optimize distributed training pipelines with PyTorch Distributed.
• Implement kernel fusion, sharding, and tiling to reduce step time.

Skills

Python

PyTorch

Data loading

Performance optimization

Analytical thinking

Education

Bachelor's/Master's/PhD in CS/CE or related

Tools

Nsight

PyTorch Profiler

CUDA

Triton

Job description

Jobtailor in Massachusetts seeks an experienced ML engineer to optimize training and inference across large models. You will profile data loading and kernel performance, implement fusion, tiling, and sharding, and collaborate with distributed training teams.

Candidates should excel in Python and PyTorch, have deep ML fundamentals, and demonstrate strong analytical problem-solving with a bias for action. Experience with CUDA or Triton is preferred, with a track record of throughput improvements.