Yifan Sun’s research focuses on large-scale optimization algorithms that arise in machine learning and scientific computing applications. This domain often includes problems that are often nonsmooth and nonconvex, in the context of heavy computation and memory requirements that must be curtailed through sampling and distribution. The research has three main themes: leveraging foundational concepts to provide intuition across applications and heuristics, developing and improving algorithms to adapt to modern day needs and resources, and giving theoretical guarantees on best- and worst-case behavior for the resulting methods.