Stellarator reactor design and optimization requires quick and accurate neutronic simulations for obtaining tritium breeding ratio, coil neutron flux, and volumetric neutron heating. In this work, an automated workflow capable of fast neutronic simulations of parametric stellarator blankets is introduced, tested, and applied. First, a differentiable parametric stellarator geometry generation package is presented that can quickly create and mesh non-uniform layered blanket geometries. Then, an existing deterministic code is improved by adding stellarator symmetric boundary conditions and adjoint methods capable of automated variance reduction. These variance reduction methods are shown to provide two order of magnitudes improvement in the figure-of-merit of a Monte-Carlo blanket simulation. The deterministic model of the workflow is then benchmarked in the parametric blanket stellarator geometry using both uniform and non-uniform blankets. Good agreement between the deterministic code and a Monte-Carlo code is obtained for tritium breeding, coil flux and volumetric heating, with the deterministic code using significantly less computational time. Finally, a simulation including the novel stellarator symmetry boundary conditions shows that a simple layered non-uniform blanket in the SQuID geometry can satisfy the neutronic engineering constraints.