Academic growth, universities have started implementing various monitoring and support systems. Many universities have implemented learning management systems (LMS) that provide a centralized platform for monitoring student performance and identifying areas where additional support may be needed. This research aims to remove a barrier between the lecturers and the students by analyzing their performance and guiding them towards achieving their goals by assigning the works and projects in the areas which a student finds difficult to deal with. This allows them to intervene early and provide the necessary support to help students succeed. The study utilized a dataset consisting of various educational features such as student demographics, parental education level, school infrastructure, and teaching quality. The dataset also included students’ performance indicators, such as grades and test scores. This dataset was collected from Kaggle. An XGBoost model along with hyper parameter optimization is being proposed to predict the students’ performance at an early stage and to minimize the failure rate. The result of this model is compared to several traditional machine learning models such as Naive bayes, Decision tree, Cumulative prediction, and Random Forest. Different metrics were used to evaluate the quality of the model and it has been observed that the proposed model outperformed by obtaining an accuracy of 98.42% compared to other models. Thus, by early monitoring of the students’ performance the lecturers could provide assessment in the lacking areas thereby mitigating the risk of students’ failures, improving the success rate and the placement ratios.