<p>Mock interviews play a crucial part in improving employability, particularly by improving effective communication—a fundamental skill in job interviews. While technical knowledge is still essential, the ability to explain ideas effectively and confidently is also critical, especially for undergraduate students from the countryside who may have had little experience to formal interview settings. As the competition for placements intensifies, mock interviews provide a valuable opportunity for students to practice, receive feedback, and build the confidence needed to succeed in real-world interviews. This study assessed student performance in mock interviews by evaluating verbal and non-verbal communication, technical skills, and overall interview readiness. Existing datasets, rubrics, and evaluation techniques used in mock interview assessments were analyzed, revealing their limitations in providing a holistic evaluation. To address these gaps, a new dataset was developed, incorporating verbal, non-verbal, technical, and non-technical skills. Features were extracted from interview videos, and regression models, including Random Forest Regressor, Support Vector Regressor, and Ridge Regressor, were applied to predict overall student performance. The results demonstrated that the newly developed dataset provided a more comprehensive assessment than existing benchmark datasets. Among the applied models, Random Forest Regressor exhibited the highest predictive capability. These findings underscored the importance of integrating diverse assessment criteria for more accurate interview evaluations. Future work should refine assessment criteria and explore advanced machine learning techniques to enhance prediction accuracy and feedback effectiveness.</p>