Alcruiter: A Proof-of-Concept for AI-Powered Resume Screening Using Large Language Models

Abstract

The first stage of talent acquisition, resume screening, suffers from inefficiency and personal bias, as recruiters manually go through large numbers of applications. This paper introduces Alcruiter, a proof-of-concept web application designed to test the idea of automating this task with Artificial Intelligence (AI). The system uses Google's Gemini generative Large Language Model (LLM) to analyze a single resume in relation to a specific job description. Built with the Streamlit framework, the application has a simple interface for job description input and resume upload, supporting PDF and DOCX formats. The main component, in the Talent Evaluation Agent class, uses precise prompt engineering to produce a numerical score and feedback. Testing results show that the LLM can consistently provide detailed, relevant, and organized evaluations, effectively identifying candidates who are a good fit versus those who are not. This project successfully proves the technical viability and potential of LLMs as an impartial, automated tool for the initial resume screening process, laying the groundwork for more advanced features.