<p>The presence of minerals in the clayey soil helps to understand the expansive behavior and the compressive strength, which are crucial for the design of the foundation. Estimation of both the properties of soil using conventional methods is costly and time-consuming. This study aims to predict these properties based on index properties of soil by leveraging artificial intelligence (AI). Bentonite and China clay are artificially mixed with varying proportions to prepare the twenty-one soil samples. X-ray diffraction analysis ascertained the mineral composition of soil samples and confirmed the montmorillonite and kaolin content. Laboratory experiments were conducted to determine index and engineering characteristics. Four AI models were trained using the laboratory test result data. For optimization, two algorithms, Levenberg Marquardt (LM) and Scaled Conjugate Gradient (SCG), along with two sets of different input and output parameters, were used. The first models of both algorithms (LM1 and SCG1) forecast the mineral composition and strength of soil, whereas the second model (LM2 and SCG2) considers the shear strength. The R² values of 0.9997 and 0.9984 for LM1 and LM2 state that LM outperformed better than the SCG. The same models were confirmed with 95% accuracy when validated against 9 natural soil samples using absolute percentage error. This research emphasizes the potential of AI in the application areas of geotechnical engineering and the exigency of validated models before implementing them in practice. The proposed model is a robust, time-efficient, and validated ANN-LM capable of accurately determining soil mineral content and strength around the world.</p>