Generic approach for spark to reuse existing application business logic

Suppose we have an application like say Twitter. The application running is built from compiling a series of classes respecting the company's business logic (henceforth application logic for referencing)

Usually, when a spark job is written say to follow or unfollow, a user, the process of following or unfollowing is duplicated in the spark environment, the table modified from spark etc.

My question: Is there a way for spark to re-use the application logic without duplicating that logic in spark? Something like an existing pattern or work-around? Any help?