In Geotechnical Engineering, the evaluation of ultimate bearing capacity emerges as an indispensable parameter, demanding meticulous assessment. With the rising popularity of machine learning (ML) techniques, utilizing ML models may offer a potential approach to addressing diverse challenges in this field. This paper introduces the use of an ML model to predict the ultimate bearing capacity (qu) of square footings of different sizes resting on clayey soil subjected to vertical load. A total of 254 experimental data, derived from load tests, were used to develop the ML model. The analysis employed a random forest (RF) algorithm to forecast the ultimate bearing capacity, showcasing its potential as a predictive tool in this domain. The important input features considered in the model included footing size (B), shear strength parameters (C & Φ) to assess the ability of the ML model. The model has been trained on a 75:25 split of data, producing predicted outcomes closely aligned with the results given by the conventional methods. R-Squared value of 0.99 on the test set and 0.98 after cross-validation has been observed in random forest (RF), demonstrating its superior performance. In conclusion, ML techniques can accurately predict the ultimate bearing capacity (UBC) of clayey soil for various square footing sizes. Based on the performance, it could be recommended that the ML model be a cost and time-effective alternative to traditional methods for determining the UBC of square footing resting on clayey soil, saving resources and valuable time without compromising reliability.