Estimating that between 10^20 and 10^60 chemical compounds could serve as small-molecule drugs, researchers face an insurmountable challenge in evaluating each one experimentally. To overcome this, MIT Associate Professor Connor Coley is developing artificial intelligence models designed to understand chemical principles, predict reaction pathways, and identify viable drug candidates. Coley, who holds joint appointments in Chemical Engineering and Electrical Engineering and Computer Science, leads a lab dedicated to applying computational methods to organic chemistry, with a primary focus on accelerating drug discovery. Coley's academic journey began in Dublin, Ohio, where he graduated high school at 16. He attended Caltech, choosing chemical engineering to merge his passions for science and mathematics. During his undergraduate years, he gained early experience in computer science within a structural biology lab using Fortran. In 2014, he joined MIT for his PhD under advisors Klavs Jensen and William Green, focusing on optimizing automated chemical reactions. His doctoral work combined machine learning with cheminformatics to design reaction pathways and develop hardware for automated synthesis, supported in part by a DARPA-funded program. This research marked his entry into using models to understand chemical possibilities. At age 25, Coley accepted a faculty position at MIT, despite mixed advice regarding returning to his alma mater. He cited the university's unique resources, interdepartmental fluidity, and vibrant ecosystem for AI and science as decisive factors. He deferred his appointment for a year to complete a postdoc at the Broad Institute, where he gained specialized experience in chemical biology and identified potential small molecules from vast DNA-encoded libraries for treating diseases. Since returning to MIT in 2020, Coley has built a lab focused on deploying AI to synthesize existing therapeutic compounds and design entirely new molecules with desirable properties. His team employs various computational approaches to pair specific chemical challenges with tailored AI solutions. One notable model, ShEPhERD, evaluates potential drug molecules based on their three-dimensional interaction with target proteins. This tool is currently utilized by pharmaceutical companies to aid in new drug discovery. Another significant project is FlowER, a generative AI model that predicts reaction products. Unlike standard models, FlowER is explicitly grounded in fundamental physical principles, such as the conservation of mass, and is constrained to consider the feasibility of intermediate steps in reaction mechanisms. This approach mimics the intuitive reasoning of expert chemists, significantly improving the accuracy of predictions regarding how reactions evolve. The lab also explores areas such as computer-aided structure elucidation, laboratory automation, and optimal experimental design. Through these diverse research threads, Coley aims to advance the frontier of AI in chemistry, ensuring that machine learning models are not just data processors but tools that understand the fundamental rules governing chemical reactions.