A typical neural network makes predictions or classifies inputs based on parameters learned from training. The accuracy of the prediction often relies on the amount of data used in training. In physical scenarios, the physics involved can be described using governing equations/PDEs. Physics-informed neural networks(PINNs) allow physical knowledge in the form of governing PDEs to be infused into the neural networks. It can make predictions without training data, although the presence of training data can improve results as well. This research targets the application of PINNs in inverse multiphase flow problems. The objective is to predict velocity, pressure and interface position at given space time coordinates. Interface position/Volume of Fluid (VOF) field is used as an auxiliary variable. Using PINNs, continuous velocity and pressure fields can be predicted by coupling the data on the interface position with the knowledge of governing physical laws. The concept of Distributed Learning Machines is being applied to improve the predictions made by PINN. Distributed learning machines are inspired by finite volume methods where the computational domain is partitioned into multiple cells, and the governing equations are solved for each cell. The solutions of these individual cells are stitched together with additional convective and diffusive flux conditions at the cell interfaces. A similar strategy is employed in Distributed PINN (DPINN).