Research Scientist / Engineer – Performance Optimization

Luma
Redwood City
Workplace: HybridFull timeFunction: Data Science & Machine LearningSkills: []

Profile and optimize Luma’s multimodal models to maximize GPU/CPU/accelerator utilization and minimize latency. Write high-performance PyTorch, Triton, and CUDA (including fused kernels and custom operations), and improve transformer implementations for distributed multi-node production deployment. Build monitoring and automation to prevent regressions, and research cutting-edge optimization techniques for transformer models using profiling tools like NVIDIA Nsight and torch profiler.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
Luma
Luma
1 week ago

Research Scientist / Engineer – Performance Optimization

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 4 hours agoStatus: Live

Job Summary

Profile and optimize Luma’s multimodal models to maximize GPU/CPU/accelerator utilization and minimize latency. Write high-performance PyTorch, Triton, and CUDA (including fused kernels and custom operations), and improve transformer implementations for distributed multi-node production deployment. Build monitoring and automation to prevent regressions, and research cutting-edge optimization techniques for transformer models using profiling tools like NVIDIA Nsight and torch profiler.
Location: Redwood City
Workplace: Hybrid
Employment Type: Full time
Job Function: Data Science & Machine Learning

Key Responsibilities

  • •Profile and optimize GPU/CPU/accelerator code to maximize utilization and minimize latency.
  • •Write high-performance PyTorch, Triton, and CUDA, including custom operations when needed.
  • •Develop fused kernels and leverage tensor cores and modern hardware features across platforms.
  • •Optimize transformer model architectures and implementations for distributed multi-node production deployment.
  • •Build performance monitoring, analysis tools, and automation, plus research new optimization techniques for transformer models.

Key Requirements

  • •Expert-level Triton/CUDA programming with strong GPU optimization skills.
  • •Strong PyTorch skills, including kernel development and custom operations.
  • •Proficiency with profiling tools such as NVIDIA Nsight and torch profiler.
  • •Deep understanding of transformer architectures and attention mechanisms.
Tech Stack:GPU optimizationGPUCPUAcceleratorPyTorchTritonCUDAFused kernelsTensor coresNVIDIA NsightTorch profilerTorch.compileTensorRTONNXXLADistributed multi-nodeTransformer architecturesAttention mechanismsWarp-level intrinsics

Company Brief

Luma
Develops AI-powered tools for capturing, editing, and rendering high-quality 3D scenes from photos and videos, enabling creators to generate photorealistic 3D assets and spatial experiences.
Industry: AR/VR & Spatial Computing
Website