We will investigate Brownian motion in particle-fluid mixtures using computational fluid dynamics (CFD)-based numerical methods. We have successfully created a CFD code that can simulate Brownian motion at different particle-fluid density ratios for a single spherical particle using previous iterations of this storage project. We are now working on extending the method to more complex particle shapes and domains. In pursuit of this goal, we will be using this project to generate memory kernel libraries for various particle-domain combinations. The generated kernel libraries will be utilised to train machine learning models, which will finally be utilised to create accurate simulations of microfluidic devices while incorporating accurate Brownian dynamics. We will be using an Immersed Boundary (IB) method implemented using IBUFlow software (written in Python) developed by Fraunhofer Chalmers Center.