Engineer on the ML team optimising model training and inference performance, including large-scale training and low-latency real-time inference. Work spans CUDA, storage, networking, and host/GPU-level considerations. Requires deep GPU and systems knowledge.