With the widespread proliferation of machine learning, there arises both the opportunity for societal benefit as well as the risk of harm. This problem requires sociotechnically grounded approaches to thoughtfully confront. However, this is hard as technical approaches tend to overfit to mathematical definitions of fairness which may correlate poorly to real-world constructs of fairness, and normative approaches may be too abstract to translate well into practice. In my research I re-orient technical ML research by interrogating the neglected normative concerns behind canonical assumptions, ground technical work in disciplines outside of computer science like psychology that have a long history of studying questions of inequality and harm, and confront practical issues that arise in the research to reality gap. I thus use my technical expertise to engage with the work of computer scientists, social scientists, and real world practitioners to develop realistic, equitable, and impactful ML fairness interventions.