Automated Resume Screening and Candidate Ranking System Using BERT-based Contextual Embeddings

In the modern recruitment landscape, Human Resource (HR) departments are inundated with a high volume of job applications, making manual screening a time-consuming and error-prone process. Traditional keyword-based screening methods often fail to capture the semantic context of a candidate's qualifications, leading to the rejection of suitable candidates or the shortlisting of irrelevant ones. This paper proposes an automated Resume Screening and Ranking System leveraging Natural Language Processing (NLP) and Deep Learning techniques. Specifically, the system utilizes Bidirectional Encoder Representations from Transformers (BERT) to generate contextual embeddings for both resumes and job descriptions (JDs). By calculating the cosine similarity between these embeddings, the system ranks candidates based on semantic relevance rather than mere keyword overlap. Experimental results demonstrate that the proposed system achieves superior precision and recall compared to traditional TF-IDF and Word2Vec approaches, significantly reducing the time-to-hire while maintaining high screening quality.