Ittai Rubinstein

Ittai Rubinstein is a graduate student in electrical engineering and computer science, developing tools to ensure the robustness and reliability of machine learning models trained on real-world data. His research addresses the challenge of data attribution—understanding how changes in a training set can impact a model’s output—by building algorithms that offer strong theoretical guarantees and practical efficiency, even in high-dimensional regimes. As a MathWorks Fellow, Ittai will advance new techniques, such as higher-order corrections to influence functions, designed to enhance the accuracy of model sensitivity analysis, and develop faster methods for uncertainty quantification. He will use MATLAB’s linear algebra libraries, visualization capabilities, and flexible interface to implement and test these tools across large datasets. Ittai’s work could improve confidence in statistical inferences across fields such as economics, public policy, and healthcare, empowering researchers to audit and interpret their models with greater precision and transparency.