This paper explores the attempt to re-engineer biological functions using a model-free reinforcement learning approach. We consider the problem of control of a genetic regulatory network. The control problem was defined as controlling the levels and dynamics of one or both genes in a toggle switch toy system to achieve feedback control. This problem aims to drive the concentrations of two specific proteins to a target region in the state space. The algorithm leverages reinforcement learning and control theory to guide synthetic biology, meaning there is no need for a complete understanding of the biological system. Our approach consists of adapting a MATLAB model and developing it using reinforcement learning. To illustrate the application of the RL model-free algorithm, a few different iterations of the algorithm are tested. The outcome of the experiments suggests that this approach can be feasible, and while it depends on preference, the DQN approach outperforms them all. The results demonstrate the capabilities of the model-free RL algorithm on a simulated data set and that it can be adaptable to a wide range of biological systems due to its ability to be efficiently reworked and utilised. Finally, wenwill discuss the RL approach's limitations, the project's limitations, and what could be done in the future.