Additive manufacturing (AM) of high-performance alloys has gained increasing attention due to its capability to produce complex geometries with tailored properties. In this study, Inconel 625 powder is processed using the laser powder bed fusion (LPBF) AM technique to fabricate tensile specimens. The printed samples are first characterized through electron backscatter diffraction (EBSD) to investigate their microstructural features, EBSD provides grain orientations, crystallographic textures, Taylor factors, and geometrically necessary dislocation densities, all of which play a critical role in the mechanical response of the alloy. Subsequently, tensile testing is performed to obtain the corresponding stress-strain curve of the samples. Building on these experimental results, a constitutive model is developed to establish correlations between the model’s unknown parameters and the microstructural data obtained from EBSD. The novelty of this work lies in the formulation of the constitutive equation, by analyzing parameters that are traditionally treated as constants, such as hardening coefficients or strain-rate sensitivities, etc. to be expressed as functions of EBSD-derived microstructural descriptors. The aim is to simulate the tensile behavior of the LPBF-manufactured specimens with improved accuracy by directly linking mechanical performance to microstructural characteristics. Furthermore, the study explores the possibility of correlating process parameters with EBSD-derived features. If such a relationship can be identified, it would enable the prediction of mechanical behavior directly from process parameters, reducing the need for extensive experimental testing. Based on the results, the proposed model demonstrates strong predictive capability for capturing the tensile behavior of LPBF-fabricated Inconel 625 directly from microstructural features. By redefining constitutive parameters as functions of EBSD-derived descriptors, the model achieves accurate stress–strain predictions with minimal computational cost. Compared to traditional approaches, the framework reduces error in yield strength and strain hardening predictions by an anticipated margin of 5-10%, while significantly improving correlation between simulated and experimental curves.