Performance Engineer
AI
CuspAI is seeking a Performance Engineer to optimize large-scale AI training and inference workloads on high-performance computing infrastructure. The role focuses on profiling, GPU kernel optimization, and distributed training efficiency to accelerate materials discovery. Requires 3+ years of experience in performance engineering or HPC/AI workloads.
What you'll do
- Analyze model architectures to optimize how mathematical operations map onto hardware, specifically focusing on NVIDIA GPU kernels and collective communication layers (NCCL).
- Implement state-of-the-art efficiency techniques such as quantization, model pruning, and distillation to lower latency for materials “search engine” APIs.
- Streamline data pipelines to ensure high-performance accelerators are never “starved” for data during massive training runs.
What you need
Must-have
GPU architectureCUDATritonPyTorch DistributedDeepSpeedMegatron-LMNsight SystemsNsight Compute
Nice-to-have
TPUsMLOpsAWSGCPAzureMaterials ScienceChemistryPhysics
Soft skills
Builder mindsetAutonomyCollaborationInterdisciplinary teamwork
Performance EngineerAIMaterials ScienceHPCGPUCUDADistributed TrainingProfilingOptimizationStartup