In modern education, traditional manual grading of subjective assessments is often time-consuming, inconsistent, and prone to bias, posing challenges for both educators and students. The “Enhanced Academic Assessments Using BERT and Top Sampling for Student Performance Optimization” project addresses inefficiencies in traditional grading, which is time-consuming and biased. This model automates subjective exam grading using BERT to ensure fair and consistent evaluation, eliminating human biases. The system also uses top sampling to generate personalized quizzes tailored to each student’s performance, supporting adaptive learning by addressing individual knowledge gaps. This AI-driven approach reduces educator workload, delivers timely feedback, and enhances the overall learning experience, fostering improved academic outcomes through more efficient, consistent, and personalized assessments.