Authors: Punit Chauhan, Aakash Chouhan, Sunny Maurya, Siddhesh Mundhe, Prof. Shilpa Doke
Abstract: In today's highly competitive job market, candidates often struggle to optimize their resumes for Applicant Tracking Systems (ATS) and lack access to realistic interview preparation environments. This paper presents ResuMentor, a full-stack, AI-driven web platform designed to bridge this gap by providing intelligent resume analysis and real-time mock interview simulation. The system accepts user-uploaded resumes in PDF or DOCX format alongside a specified job role or description, and leverages the OpenAI GPT-4o API via Spring AI to generate ATS compatibility scores, keyword gap analysis, and actionable improvement suggestions tailored to the target job profile. For interview preparation, ResuMentor deploys an AI voice agent that conducts a structured, 30-minute mock interview session, dynamically generating questions ranging from beginner to advanced level based on the parsed resume content. The platform employs the Web Speech API for real-time speech-to-text transcription, providing a live transcript visible to the user during the session. Post-session, a detailed feedback report evaluates the clarity, conciseness, and relevance of the candidate's responses with specific examples drawn from the transcript. The backend is developed using Java Spring Boot 3.3 with Spring Security and OAuth2 for Google-authenticated login, MySQL as the relational database, and Apache Tika for resume parsing. The frontend is built with plain HTML, CSS, and JavaScript, featuring a responsive dark/light theme toggle. A personalized dashboard tracks historical ATS scores and interview performance trends using Chart.js visualizations, enabling users to monitor their growth over time. ResuMentor demonstrates that integrating large language models into career development tools can significantly improve candidate preparedness and resume quality.
