Quick overview
Job description
About the Role
Join an AI company building systems for training and serving image, video, and world-model workloads. You will improve performance across the stack, from GPU kernels to distributed training and inference engines.
What You'll Do
Optimize training and inference performance for image, video, and world-model workloads.
Profile and resolve bottlenecks at the kernel, memory, system, and cluster levels using Nsight and related tools.
Implement low-level optimizations with CUDA and Triton.
Design and improve distributed training and inference engines for diffusion models.
Improve communication performance across GPUs and nodes using technologies such as NCCL, RDMA, InfiniBand, and RoCE.
Build benchmarks and regression tests to preserve performance improvements in production.
Collaborate on hardware-aware kernel, runtime, and model design.
What We're Looking For
Two or more years of experience in deep learning training or inference systems, distributed systems, or related ML systems work.
At least one year of recent, hands-on experience in ML systems or GPU performance engineering.
Experience authoring core features in an inference or training framework, such as vLLM, SGLang, TensorRT-LLM, or Megatron.
Hands-on experience writing kernels for NVIDIA GPUs using CUDA, CUTLASS, Triton, or PTX.
Strong machine learning and computer science fundamentals, with a degree in computer science or a related quantitative field.
Experience optimizing inference or training performance across GPU kernels, runtimes, or distributed execution.
Experience with diffusion, video, image, or other multimodal workloads is relevant.
Compensation & Benefits
Base salary: $180,000 to $260,000 annually. Visa sponsorship is available.
Location
On-site in Menlo Park, California, United States.