Additive Manufacturing (AM) offers a sustainable approach to building parts layer-by-layer from 3D models. AM has demonstrated a promising potential in reducing material and energy consumption, shortening supply chains, etc. However, current environmental assessments of AM often rely on the knowledge and resource-intensive Life Cycle Assessment (LCA) methodology, and often consider a limited set of design and process parameters. Thus, selecting optimal design and process parameters can be challenging, particularly during early design-evaluation cycles. To expedite design iterations and reduce the knowledge demands of LCA modeling, developing an automated LCA tool could mark a remarkable leap forward in further improving AM’s sustainability. One promising avenue involves integrating Machine Learning (ML) with LCA, the effectiveness of which has been demonstrated outside the AM domain. This thesis offers various contributions to the research community. Firstly, it identifies a comprehensive set of influential AM design and process parameters affecting AM’s sustainability and summarizes their implications. Secondly, it proposes a novel framework that combines LCA, ML, and product-process co-design to automate AM's environmental assessment, featuring a pioneering step in the field. The framework's effectiveness is demonstrated through the creation of a training dataset, refinement of design-process parameters, and the utilization of various ML algorithms to develop a data-driven LCA (DD-LCA) model for the Fused Filament Fabrication (FFF) process. Notably, Extreme Gradient Boosting (XGBoost) and Multi-Layer Perceptron (MLP) demonstrate impressive prediction accuracy of 98% and a root mean square error (RMSE) of 0.029. The model exhibits robust generalizability when tested on new data. By bridging the gap between LCA, ML, and product-process co-design in the context of AM, the model is anticipated to mitigate the challenges and limitations associated with the current LCA-based assessment tools. Lastly, this thesis employs an evolutionary-optimization algorithm, the Non-dominated Sorting Genetic Algorithm II (NSGA-II), to create an AM sustainability optimization tool without compromising quality. To facilitate diverse multicriteria decision-making, the Euclidean distance and the Technique of Order of Preference by Similarity to Ideal Solution (TOPSIS) are utilized to find the best optimal of all received Pareto solutions. This tool elevates sustainability by replacing the current "fix it later" approach.