The mammalian retina tissue is positioned at the rear of the eye and comprises 10 layers, which include 6 different neural sub-types. Many studies have investigated the development of these cell types and their laminated structure, but current computational models are limited. Here, we use agent-based modelling to create a computational model involving gene regulatory mechanisms to reproduce the retinal layer architecture from a single progenitor pool. The progression of a progenitor pool into a functional biological structure comprised of multiple cell types is orchestrated by a series of stochastic fate determining events. The dynamicity of the gene regulation that controls this process means the probability of any given restrictive event is relatively volatile. To address this challenge, our computational gene regulatory network describes these probabilities on the basis that they are continuously changing, so to more closely reflect the biological rate of which specialised cell types emerge. Implementing this concept into the powerful agent-based modelling software BioDynaMo provides a way to study the effect of these stochastic model parameters. Defining these variables in an agent-based model shows us their influence on specialised cell count and how their properties culminate to form a laminated framework in a 3D-space. We also incorporate characteristics of the specific cell types, which initiates different cellular behaviours as the specialised cells arise. This allows them to interact with each other and their environment, where they are influenced by both internal and external cues that direct them to form their characteristic structure. By incorporating machine learning techniques, we also harness the use of running multiple simulations in parallel to fine-tune model parameters to identify suitable parameter combinations that accurately reflect biological observations.