Senior ML Training Systems Engineer - Distributed CUDA

Genesis AI; San Francisco (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Genesis AI in San Francisco is seeking a senior ML infrastructure engineer to design and optimize distributed training systems and performance-critical components. You will profile bottlenecks, implement low‑level code (CUDA, Triton) and ensure efficient hardware utilization across multi‑node GPU clusters.

Join a team focused on scalable AI foundations, monitoring tools, and robust performance improvements for large‑scale runs.

Qualifications

• 8+ years of distributed systems, ML infrastructure, or high-performance computing experience.
• Production‑grade Python programming expertise.
• Low‑level performance mastery: CUDA/cuDNN/Triton and CPU‑GPU interactions.
• Experience with PyTorch training jobs and multi‑node scaling.
• Track record tuning hardware‑software interactions for maximum utilization.

Responsibilities

• Profile bottlenecks across the foundation model training stack and eliminate them.
• Design, build, and optimize distributed training systems for multi‑node GPU clusters.
• Implement low‑level code (CUDA/cuDNN/Triton) and integrate into high‑level training frameworks.
• Optimize workloads for hardware efficiency: CPU/GPU balance, memory, throughput, and networking.
• Develop monitoring and debugging tools for large‑scale runs to diagnose performance regressions.

Skills

Distributed systems

Python

PyTorch

CUDA

cuDNN

Triton

Kernel optimization

High performance

Data pipelines

GPU utilization

Tools

PyTorch

CUDA

cuDNN

Triton

Job description

Genesis AI in San Francisco is seeking a senior ML infrastructure engineer to design and optimize distributed training systems and performance-critical components. You will profile bottlenecks, implement low‑level code (CUDA, Triton) and ensure efficient hardware utilization across multi‑node GPU clusters.

Join a team focused on scalable AI foundations, monitoring tools, and robust performance improvements for large‑scale runs.