In recent years, a significant amount of resources and attention has been directed at increasing the diversity of the hi-tech workforce in the United States. Generally speaking, the underrepresentation of minorities and women in tech has been understood as an “educational pipeline problem,” – for a variety of reasons, these groups lack the social supports and resources needed to develop marketable technical literacies. In this thesis I complicate the educational pipeline narrative by taking a close look at the perspectives and practices of three different groups. First, I explore widespread assumptions and recruitment practices found in the tech industry, based on interviews I conducted with over a dozen leaders and founders of tech companies. I found that widespread notions of what merit looks like (in terms of prior work experience and educational pedigree) have given rise to insular hiring practices in tech. Second, I offer an in-depth examination of the risks and opportunities related to an emerging set of practices termed “algorithmic recruitment,” which combines machine learning with big data sets in order to evaluate technical talent. Finally, I analyze the strategies adopted by a non-profit called CODE2040 in order to facilitate structural changes in how tech recruits talent to include a more diverse set of qualified applicants. I conclude by offering a more robust conceptualization of diversity and its value in the tech sector, as well as some specific ways to increase tech’s diversity in the future.