Estimating Above-Ground Biomass (AGB) is crucial for evaluating carbon cycles and monitoring the vitality and productivity of forest ecosystems. Conventionally, allometric equations have been employed for AGB estimation, valued for their straightforward approach and reasonable precision. Estimating Above Ground Biomass (AGB) is pivotal in understanding carbon dynamics and assessing the health and productivity of forest ecosystems. Traditionally, allometric models have been widely used for AGB estimation due to their simplicity and relative accuracy. However, these models often suffer from limitations, including species specificity, geographic constraints, and the assumption of fixed nonlinear relationships between AGB and independent variables. Such constraints highlight the need for more flexible and generalizable approaches. Currently, machine learning (ML) has emerged as a powerful tool for AGB estimation, offering significant advantages over traditional methods. ML algorithms excel at handling complex, nonlinear relationships, and accommodating mixed data types. This chapter explores the application of various ML models for estimating AGB in diverse forest ecosystems. In the current study, field data on Long Leaf Mahogany (Swietenia macrophylla) on 650 trees was collected from 41 farmer’s fields in 8 districts in Maharashtra and one in Gujrat, India. Using diameter at breast height (DBH) and tree height (H) as independent variables, the ML models were trained and validated through a rigorous process, including hyperparameter tuning and tenfold cross-validation. Various statistical metrics were employed to evaluate model performance. Furthermore, a weighted ensemble approach combining multiple ML models was also implemented to enhance estimation accuracy. The Weighted Average ensemble model exhibited the strongest overall performance among the evaluated models. It achieved the lowest RMSE (0.025) and MAD (0.009), indicating minimal prediction error. Additionally, the ensemble model attained the highest R2 value (0.976), reflecting its superior ability to explain variability in the data. Among the individual models, RF, XgBoost, and MARS achieved high R2 values exceeding 0.9, indicating strong predictive capability. The results underscore the potential of ML models to provide more precise and scalable solutions for AGB estimation. contributing to more informed decision-making in forest management and carbon accounting.