Over the last thirty years, computer modeling has evolved into an essential, widespread activity in drug discovery. The contributions of our group have been in multiple areas including virtual screening, prediction of ADME properties, and FEP-guided lead optimization. The advances were all accompanied by development of new software including the BOMB program for de novo design, QikProp for AI-based estimation of pharmacological properties and metabolites, and MCPRO for Monte Carlo simulations of protein-ligand complexes.   To pursue prospective studies, our efforts were expanded to include synthesis, biological assaying, and crystallography. Emphasis has been placed on optimization of the hits from virtual screening to yield potent, drug-like inhibitors. MC/FEP simulations are used to identify the most promising choices for substituents on rings, heterocycles, and linking groups.  The illustrated applications center on inhibitor design for HIV-1 reverse transcriptase and SARS-CoV-2 main protease. Micromolar leads have been rapidly advanced to low nanomolar inhibitors. Key computational issues are also addressed.