My work focuses on applying mathematical, statistical, and probabilistic tools to archaeological and palaeoanthropological datasets. Archaeology faces unique challenges, including extremely small, incomplete, and fragmented records, as well as the increasing dependence on complex “black-box” computational models whose internal logic is often opaque. In this talk, I will outline how I have been addressing these challenges through a set of complementary research programmes: from augmenting and completing palaeoanthropological datasets, to investigating the mathematical structure of the iso-parameters used in use-wear and microwear analysis. Together, these efforts illustrate how an explicit, rigorous archaeological data science can help us better understand the past—and the tools we use to study it.