Is Set Theory Engineering?

If we see mathematical sets as a linguistic plural referencing mechanism then the question of their existence falls away. Since they are artificial there remains the engineering question, Do they work reliably? In this context, Do they lead us into contradiction? The applied mathematics of plural referencing, aka Axiomatic Set Theory tries, in part, to answer this question.  But unlike other uses of the axiomatic method in mathematics, the system of ZF is not trying to capture truths about some class of entity. Rather, it's an engineering specification that we have drawn up.  For example, the pairing axiom is not making a conventional existence assertion. What it is saying is that if referents a and b exist or are themselves sets, then it's safe to use the set {a, b}. Safe in the sense of not leading to contradiction. The strange status of the Axiom of Choice now makes more sense. We know that AC is consistent with the other axioms of ZF, so we are free to use a referencing mechanism that satisfies AC or one which does not, according to choice (!)