I think the vision outlined is a great one. As professor Cottenier mentioned its probably a maturation of the field. Such that you can pull in tried and tested engineering practices similar to fluid simulations for piping design, raytracing for lense design,EMI simulations in electronics.. You would have software where you put in certain constraints in terms of material properties and it does a calculation based on some persisted info and some realtime ad hoc info.
In terms of the feasability my first reflex would be to say that a mole seems big, so solving something with potentially on the order of 10^23 elements seems rather hard. I have no idea as to the state of the endeavour and hope to learn more in this course.