Computing the fluid phase interfaces in multiphase flow is a challenging area of research in fluids. The Volume of Fluid andLevel Set methods are a few algorithms that have been developed for reconstructing the multiphase fluid flow interfaces. The thesis work focuses on exploring the ability of neural networks to reconstruct the multiphase fluid flow interfaces using a data-driven approach. The neural network model has liquid volume fraction stencils as an input, and it predicts the radius of the circle as an output of the network which represents a phase interface separating two immiscible fluids inside a fluid domain. The liquid volume fraction stencils are generated for randomly varying circle radii within a 1x1 domain using an open-source VOFI library. These datasets are used to train the neural network. Once the model is trained, the predicted circular phase interface from the neural network output is used to generate back the predicted liquid volume fraction stencils. Error norms values are calculated to assess the error in the neural network model’s predicted liquid volume fraction stencils with the actual liquid volume fraction stencils from the VOFI library. The neural network parameters are optimized by testing them for different hyper-parameters to reduce the error norms. So as to minimize the difference between the predicted and the actual liquid volume fraction stencils and errors in reconstructing the fluid phase interface geometry.