Intelligent Resume Screening and Candidate Ranking Using CNN-GRU: A Deep Learning Approach for Smarter Recruitment

Automated resume screening is a crucial task in modern recruitment, requiring efficient and accurate candidate evaluation. This research proposes a hybrid Convolutional Neural Network-Gated Recurrent Unit (CNNGRU) model for resume classification and ranking, leveraging CNN for feature extraction and GRU for sequential dependency learning. The model is trained on a dataset of resumes labeled based on job relevance, with key evaluation metrics including accuracy, precision, recall, and $\mathbf{F 1}$-score. Experimental results show that the proposed CNN-GRU model achieves 93.7% accuracy, outperforming traditional machine learning models such as Support Vector Machine (79.2%), Random Forest (82.6%), and deep learning approaches like LSTM (88.4%) and CNN (90.1%). Furthermore, the candidate ranking system successfully differentiates applicants based on experience, skills, and education, ensuring a fair and objective selection process. The histogram of ranking scores confirms that most candidates fall within the 75 %-90 % range, demonstrating the model’s ability to assess resumes effectively. These findings highlight the CNN-GRU model’s efficacy in automating recruitment, reducing manual effort, and improving decision-making accuracy. Future research may focus on expanding dataset diversity and optimizing hyperparameters to further enhance model performance across various job sectors.