Performance Engineer
Core
Building optimization layers and profiling infrastructure to maximize efficiency of large-scale AI training and inference workloads for materials discovery.
Role type
Senior IC performance engineer (HPC/AI)
Builds
High-performance compute clusters and optimization tools for AI training pipelines
Domain
AI infrastructure / High-Performance Computing (HPC)
Deliverable
production ML models
Required skills
GPU architecture optimization, CUDA kernel development, distributed training frameworks, profiling tools, system observability, data pipeline optimization
Preferred skills
TPU/alternative accelerator experience, MLOps practices, open-source contributions
Technologies
NVIDIA GPUs, NCCL, PyTorch Distributed, DeepSpeed, Megatron-LM, Nsight Systems, Nsight Compute, Triton, H100 clusters
Responsibilities
Analyze model architectures to optimize hardware mapping; Implement quantization, pruning, and distillation techniques; Build real-time dashboards for system metrics; Profile training runs to identify bottlenecks; Collaborate with researchers to adapt models for hardware compatibility.
Seniority
Senior, hands-on IC
