In the swiftly developing pharmaceutical industry, the efficiency and reliability of the tablet press equipment play a central role in ensuring continuous production and maintaining the quality of products. This paper investigates the use of machine learning models for predicting faults in tablet press machines to create a proactive maintenance system that can forestall potential failures and prevent operational halts. With a synthetically created dataset cycling the real operational parameters and past failure occurrences of a tablet press machine, proposed and assessed various machine learning models – Random Forest Classifier, Support Vector Classifier, K-Nearest Neighbors, Naive Bayes, and Gaussian Process Classifier – capable of detecting patterns indicative of imminent failures. The Random Forest Classifier posts the best results by far. The performance evaluation metrics – Accuracy, Precision, Recall, and F1-Score – indicate that the Random Forest Classifier records the best performance, correctly predicting both failure and non-failure instances. This paper ascertains that machine learning can be applied to build models that adequately predict faults and mitigate downtime and wastage associated with pharmaceutical production. It paves the way for more advanced ML-based fault prediction systems in industries.