Ti64 is widely used, particularly, in the realm of medical implants, due to its superior biocompatibility. This work seeks to establish a machine learning (ML) based approach to optimize the laser powder bed fusion (LPBF) process parameters to fabricate Ti64V medical implants. The goal is to improve the dimensional accuracy by minimizing the distortion. This is achieved by using a surrogate geometry which captures the essence of the reference implant geometry. Since the structure of interest is a trabecular bone, we have adopted a porous scaffold of Voronoi-based lattice structure as the surrogate geometry. The input LPBF process parameters viz. scan speed, hatch spacing and laser power are employed for Ansys-thermomechanical simulations. The simulation parameters and outcomes such as stress and displacements were used to develop ML models with suitable algorithms. The ML framework predicts the stress and displacement based on the input LPBF process parameters and the geometry’s spatial coordinates. Further, optimal process parameters that minimize stress and distortion are also predicted. Therefore the proposed framework has implications in terms of material and cost saving along with desired part accuracy.