In this memo, I proposed a data-driven redesign of a job training program using predictive modeling to allocate limited resources fairly and effectively. I evaluated algorithmic decision points with cost-benefit and equity tradeoffs, recommending a 51% threshold to maximize fairness and social impact. While focused on criminal justice, this project demonstrates my ability to connect data science with policy, stakeholder needs, and systemic bias, a transferable framework I also apply to transport and mobility challenges.
