The present paper introduces a novel paradigm for Physics-Informed Neural Network (PINN) based Topology optimization framework. Unlike the existing machine learning-based topology optimization, the proposed framework utilized Deep Energy Method (DEM) PINN and neural reparameterization scheme to replace Finite Element Analysis (FEA) and sensitivity analysis operations in the conventional topology optimization algorithm. The two PINNs that respect the governing physical equations in the format of Partial Differential Equations (PDEs) are trained without any labelled data [1]. In this approach, DEM-PINN is trained to numerically determine the deformation states of the given computational domain by solving overall potential energy as the natural loss of PINN [2]. Subsequent sensitivity analysis is carried out using the automatic differentiation feature in deep learning to solve the derivative of the objective function with respect to the design variable [3]. The effectiveness and capabilities of the proposed framework are assessed through several stiffness maximization problems for both two and three-dimensional case studies, highlighting its potential to advance the field of topology optimization.