In this talk, we present the StructJuMP, a parallel, memory-distributed modeling framework that matches the parallel capabilities of optimization solvers such as PIPS. StructJuMP distributes the model and the function and derivative evaluation work needed by the optimization solver by exploiting the structure of stochastic optimization problems we target, i.e., security constrained optimal power flow problems. We demonstrate StructJuMP’s intuitive syntax for modeling the problem’ structure. Some preliminary parallel execution results are also presented. In the second part of this talk, we focus on the automatic differentiation which is also used by StructJuMP for its derivative evaluations. We briefly introduce the Edge Pushing (EP) algorithm for Hessian computation, and then demonstrate that using EP can be more beneficial to the runtime performance than the Coloring algorithm which is currently used in our modeling framework.