Determining the magnetic ground state of a material — its ordering type, moment magnitudes, and the exchange interactions that stabilize it — is a central bottleneck in computational magnetism. First-principles methods based on density functional theory (DFT) can predict these properties with high accuracy, but calculating exchange parameters, magnetic anisotropy energies (MAE), and comparing competing spin configurations across the many candidate structures, dopants, and compositions of interest in magnetic materials research is computationally demanding. This project develops a combined first-principles and machine-learning (ML) framework for efficiently classifying magnetic ground states across broad classes of magnetic materials, spanning itinerant ferromagnets, antiferromagnets, and doped or chemically substituted derivatives with modified exchange topologies. The workflow proceeds in two stages. First, we perform systematic DFT calculations — including spin-orbit-coupled total-energy and MAE calculations, and pairwise exchange-coupling (Jᵢⱼ) extraction via the magnetic force theorem / Lichtenstein-Katsnelson-Antropov-Gubanov (LKAG) formalism — on a representative set of magnetic compounds and their structural or compositional variants. These calculations are carried out using VASP for total-energy and anisotropy calculations, and RSPt and SPR-KKR (a full-potential linear muffin-tin orbital code and a Korringa-Kohn-Rostoker Green's function code, respectively) for exchange-interaction calculations, chosen because each offers complementary strengths for different classes of magnetic ordering, correlation strength, and disorder treatment (including DFT+U and dynamical mean-field extensions for strongly correlated systems). Second, the resulting exchange parameters, anisotropy energies, and derived quantities (estimated ordering temperatures, spin-wave stiffness) are used as a feature set to train supervised ML classifiers (e.g., gradient-boosted trees or graph neural networks) that map structural and compositional descriptors to magnetic ordering class and approximate critical temperature. The scientific objectives are: (1) to build a curated, physically consistent database of DFT-derived exchange interactions and magnetic ground-state labels spanning multiple structural families, substitution patterns, and correlation regimes; (2) to train and validate ML models capable of predicting magnetic ground-state character directly from structure and composition, bypassing the need for a full exchange-parameter calculation on every candidate; and (3) to apply the trained models to screen large compositional and defect spaces for candidates with enhanced magnetic anisotropy, exchange coupling, or ordering temperature, prioritizing the most promising cases for full first-principles follow-up.