This tutorial package delves into an advanced inverse modeling approach for predicting carbon fiber properties in composite materials using a machine learning (ML) technique. Specifically, it covers the use of Gaussian Process Regression (GPR) to build a surrogate model for accurate predictions of fiber properties based on data from unidirectional (UD) lamina. By leveraging Finite Element (FE) homogenization, synthetic data is generated for training the GPR model, accounting for variations in fiber, matrix properties, and volume fractions. This framework’s efficiency and accuracy are validated using real-world data, highlighting its potential as a computational alternative to traditional experimental methods. The package includes detailed explanations, case studies, and practical exercises, equipping users with hands-on experience in applying this ML-based approach to composite material analysis.