Penn AI Seminar feat. Andrew Zahrt

Machine-Learning-Guided Solvent System Engineering

Our research develops enabling technology to streamline process development in organic chemistry. One such effort is Machine-Learning-Guided Solvent System Engineering (MLSSE), which uses data-driven methods to design higher-order solvent blends—an opportunity often overlooked because design principles are limited. Our workflow applies machine learning and computational analysis to identify optimal blends and quantify how mixed solvents influence reactivity, advancing the fundamental understanding of solvent effects. Practically, MLSSE aims to boost yield and selectivity, lower catalyst loadings or replace niche catalysts to improve manufacturability, and substitute hazardous solvents with greener alternatives.