Senior Research Engineer - Performance Optimization

Candidates should have significant problem-solving experience in PyTorch, CUDA, and distributed systems. They must have experience training large models using Python and PyTorch, along with practical experience working with the entire development pipeline from data processing to inference. Additionally, candidates should have experience optimizing and deploying inference workloads for throughput and latency, profiling CPU and GPU code in PyTorch, and writing high-performance parallel C++ code. Familiarity with high-performance Triton/CUDA and writing custom PyTorch kernels is also required. Good to have experience with deep learning concepts such as Transformers and multimodal generative models like Diffusion Models and GANs, as well as building inference/demo prototype code including Gradio and Docker.