Proteins often target specific areas in the cell. Some reside exclusively inside the cell, some embed themselves in the cell wall, and some are tethered to the outside of the cell. Others are actively secreted. Designing therapeutic drugs or proteins must account for the location of the target protein. Many important targets, like viral proteins, are initially inside the cell. Developing proteins that can specifically recognize and bind to them inside a cell is challenging. The cell interior presents many competing molecular interactions due to crowding, protein aggregation, and local ionic concentrations. Therapeutics designed to target a specific protein can exhibit off-target binding. This project will use a combination of computational modeling and experiments to overcome off-target binding. Tens of millions of binder variants will be screened. The resulting datasets can be used to train AI models to design intracellular protein binders more efficiently. This project will also contribute to workforce development by integrating the technology into a new undergraduate course and by providing research opportunities for local high school students. A yeast display platform will be used to overcome the limitations of traditional protein screening methods. It will enable simultaneous screening of a single binder variant under both oxidizing (extracellular) and reducing (intracellular) environments. Using partially efficient ribosome-skipping sequences, the approach will express each binder in both soluble cytoplasmic and surface displayed forms. High-throughput fluorescence-activated cell sorting will be used to quantitatively assess binding in the extracellular environment. The intracellular binding will be monitored through Forster resonance energy transfer, taking advantage of the quantum mechanical effects of energy transfer due to dipole-dipole coupling. The approach will be applied to evolving nanobodies and improving de novo designed binders through large-sca