Automation and enhancement of recruitment processes with large language models (LLM)

This study provides a comprehensive examination of using Large Language Models (LLMs) in job application processes, aiming to improve both automated job listing discovery and résumé tailoring. An analysis conducted on 2,490 resume collected from various sectors—including IT, finance, and engineering—demonstrates that the scoring and adaptation mechanism designed based on OpenAI GPT and Google Gemini APIs yields significant improvements particularly in keyword matching score (KMS) performance. An initial review of the data shows that while candidates generally meet contextual releveant score (CRS) and experience fit score (EFS) criteria, there are gaps in keyword alignment. When LLMs are employed, the marked increase in KMS rates leads to an overall rise of approximately 6–7% in resume scores. However, gains in metrics such as CRS and EFS remain relatively limited, indicating that human input is still pivotal in these areas. The proposed system relies on machine learning algorithms and feedback channels to optimize Applicant Tracking Systems (ATS), continuously refining application materials using feedback from employers and ATS rankings. Due to concerns around data privacy and legal restrictions, the Machine Learning Prediction Score (MLPS) component is temporarily deactivated but appears to have potential for reintegration with more comprehensive models in the future. Enhancing ATS compliance, increasing the precision of candidate–job matches, and speeding up the application process result in notable time and effort savings for both employers and applicants. Nonetheless, the findings highlight ethical concerns and bias management within automation. While LLM-based algorithms excel in technical tasks such as keyword matching, they still require human oversight and contribution for aspects like experience suitability and contextual analysis—complex judgments best handled by human expertise. Especially in forming fair and impartial hiring processes, the balanced and integrated use of automation and human expertise remains paramount. In this regard, ongoing research will continue to focus on bias reduction and broader adoption of automatic job application systems.