Lisa Torrey
Njeri's project examined algorithmic fairness in resource allocation by comparing synthetic scholarship allocation models with real-world mortgage lending data. Using Python, machine learning, and established fairness metrics, she evaluated how predictive models can produce different outcomes across demographic groups, even when sensitive attributes such as race and sex were excluded from the training process. By comparing multiple fairness metrics, she found that no single measure fully captures model fairness, demonstrating the importance of evaluating algorithms from multiple perspectives. Throughout the summer, Njeri gained experience working with real-world datasets, building machine learning pipelines, and interpreting fairness metrics. Her project strengthened her interest in developing accurate, transparent, and equitable technologies and reinforced her goal of pursuing graduate study in ethical artificial intelligence and algorithmic fairness.