Analyze and optimize AI training and inference workloads across distributed systems. Design benchmarks, profile performance, identify bottlenecks, and optimize communication software using MPI, NCCL, UCX, libfabric, and RDMA. Translate modeling and benchmarking results into validated system improvements while coordinating technical work across hardware, software, networking, and architecture teams.

Lead performance strategy for large-scale AI training and inference systems. Investigate workload behavior, profiling, benchmarking, modeling, distributed communication, and multi-node scaling across hardware and software. Define measurement standards, guide optimization roadmaps, review C++ and Python tools, and influence system architecture and engineering priorities across teams.