Simran Arora leads the Frontier Performance research team at Together AI and is an assistant professor of Computing and Mathematical Sciences at Caltech. Her research focuses on expanding the Pareto frontier between AI quality, memory, and compute, studying algorithms, hardware, and applications in lockstep. Professor Arora is a co-author of ThunderKittens, a widely adopted framework for writing high-performance GPU kernels, and of the Monarch Mixer and Based architectures.