Abstract: In this talk I will introduce my research on efficient multivariate variance estimators for serially correlated data. A family of partial-overlapping batch means estimators is introduced, which offers a tradeoff between computational and variance efficiency. The family is indexed by a shift parameter which determines the overlap between batches and is a generalization of non- overlapping batch means and overlapping batch means. Through the talk I aim to connect my research to my journey to statistics with two goals: clarifying abstract concepts and showcasing a non-traditional path to academia.