Performance Engineer

Skills

About the Role

You will optimize large-scale AI training and inference workloads, profile bottlenecks, and improve GPU and distributed-system performance. You will build observability dashboards, manage accelerator utilization, develop scaling tools, and collaborate on hardware-compatible model and cluster designs.

Requirements

• 3+ years of performance engineering, site reliability engineering, or low-level software development experience focused on HPC or AI workloads
• Expert knowledge of GPU architecture and CUDA kernel optimization or Triton
• Knowledge of distributed training frameworks including PyTorch Distributed, DeepSpeed, or Megatron-LM
• Knowledge of networking protocols
• Proficiency with Nsight Systems, Nsight Compute, or Weights & Biases

Responsibilities

• Optimize training and inference workloads for GPU hardware and collective communication layers
• Implement quantization, pruning, and distillation techniques
• Streamline data pipelines for large-scale training
• Profile training jobs and identify performance bottlenecks
• Build dashboards for utilization, memory bandwidth, and cluster uptime
• Monitor accelerator utilization and prevent idle time and memory leaks
• Develop scaling laws and tools for compute, memory, and networking
• Adapt model architectures for hardware compatibility
• Inform compute-cluster design using performance data

Benefits

• Equity
• 28 days holiday in DE, NL, and UK or 21 days holiday in JP, SG, and US, plus local public holidays
• 26 weeks paid parental leave for primary caregivers and 12 weeks for secondary caregivers
• Professional development budget