Role owns end-to-end modules and projects focused on high-performance communication and memory management libraries for distributed AI. Responsibilities include hardware-software co-optimization (GPUDirect RDMA, NVLink), profiling and optimizing data movement across GPU memory, DRAM, NVMe and networks, integrating networking into AI serving stacks, contributing to open-source, and mentoring. Requires 12+ years systems software/networking experience and advanced degree or equivalent.

Role involves understanding, profiling, and optimizing DL training and inference on state-of-the-art hardware; collaborating across research and engineering teams; implementing production-quality software and building automation tools. Requires 5+ years experience, strong DL and computer-architecture background, and programming skills in C++ and Python; GPU programming (CUDA/OpenCL) and LLM framework experience are advantageous.