Neutrons can be detected indirectly through a nuclear reaction where the products are ionizing radiation. Due to the shortage of 3He, several studies are searching for alternatives, where the Gas Electron Multiplier (GEM) detector using a layer of 10B as a neutron converter is a very promising option. The GEM detectors are a type of Micro-Pattern Gaseous Detectors (MPGD), widely used in particle tracking systems, as the Time Projection Chamber of the ALICE experiment in the LHC-CERN, and proposed for many other applications, including neutron detection. A common strategy to simulate this kind of detector is based on two frameworks: Geant4 and Garfield++. Given the high ionizing power of these nuclear reaction products, a full simulation is very time consuming then a fast simulator was developed using a parameterization strategy based on these two frameworks that allowed to generate enough data to study an optimized version of this detector. In this proceeding, it is shown the optimization that improves the position resolution by changing the gas mixture and/or its pressure.