Nowadays, academic institutions conduct studies to attain quality and excellence in student academic performance through Data Mining tools. This paper explores various classification techniques in predicting graduates' career specialization. The data sets used were obtained from Bulacan State University Sarmiento Campus' Information Technology graduates from 2013 to 2016. From these data, a model was created using Naïve Bayes, J48, Random Forest, and Support Vector Machine classification algorithm, with 18 attributes. Among the models built, Naïve Bayes and Random Forest algorithm yielded better accuracy rating, and acceptable ROC and RMSE values. Performance of students per subject area was also determined, and based on this; students perform satisfactorily in both soft and technical skills manifested in the highly satisfactory performance on the Internship course. However, there were few graduates who pursue a career in Networking, and measures must be undertaken to elevate performance in early Programming and Networking courses..