Benchmarking State-of-the-Art Transformer Models for Automated Resume Screening in AI-Driven Recruitment

The increasing number of job applications received by organizations has made manual resume screening a time-consuming and resource-intensive process. Transformer-based language models have emerged as a promising solution for automating recruitment tasks by understanding contextual information and semantic relationships within resumes and job descriptions. This study benchmarks several state-of-the-art transformer models, including BERT, RoBERTa, DistilBERT, and DeBERTa, using publicly available recruitment datasets to evaluate their effectiveness in automated resume screening. A qualitative research approach based on secondary data analysis is employed to compare the models across multiple performance indicators reported in previous studies, such as accuracy, precision, recall, F1-score, computational efficiency, and inference time. The findings indicate that while larger transformer models generally achieve higher predictive performance, lightweight models provide competitive results with lower computational costs. The study highlights the strengths and limitations of each model and proposes recommendations for selecting suitable transformer architectures in recruitment systems. The research contributes to the growing field of AI-driven human resource management by providing a comprehensive comparative analysis.