Computer-aided drug design (CADD) techniques are a routine part of modern drug development. Modeling before laboratory experiments maximizes the potential to discover a molecule with the best therapeutic potential among the vast number of possibilities. However, classical mechanical and quantum mechanical calculations of 3D shape and electronic structure are often too slow to be applied at a large scale. Adding machine learning to CADD workflows extends their capability to potentially operate at the scale of billions of molecules. The combination of CADD and machine learning could one day speed resource-intensive free energy perturbation (FEP) calculations that predict binding affinity with accuracy comparable to experimental measurements.