My primary research interests lie in mediation analysis, particularly in developing adaptive tests for composite mediation effects in challenging settings such as high-dimensional mediators, the presence of unmeasured confounding, and nonlinear outcome or mediator models. I am also interested in independence testing/conditional independence testing/conditional mean independence testing and its generalization to detecting conditional average treatment effects (CATE). In addition, I also focus on high-dimensional statistical inference and causal machine learning.