AI-Powered Virtual Interviewer with Real-Time Feedback

The interview process is a critical gateway to professional employment, yet it remains heavily reliant on subjective human evaluation, scheduling constraints, and inconsistent assessment criteria. Candidates from diverse backgrounds often lack access to structured practice environments that can simulate real interview scenarios and deliver actionable, personalized feedback. This paper proposes an AI-Powered Virtual Interviewer system that addresses these limitations by delivering an end-to-end, automated interview experience with real-time, multi-dimensional feedback. The proposed system dynamically generates domain-specific questions using a fine-tuned large language model, transcribes candidate responses through an advanced automatic speech recognition module, evaluates answer quality through semantic similarity scoring, and simultaneously analyzes communication cues including speech fluency, sentiment, and facial expressions. A personalized post-interview feedback report is automatically generated and delivered to the candidate. This paper surveys five closely related research works in automated interview analysis, spoken dialogue systems, virtual coaching agents, and LLM-based question generation, identifies persistent gaps in existing approaches, and describes how the proposed system is designed to close them.

Keywords - Automated Interview System; Large Language Models; Automatic Speech Recognition; Real-Time Feedback; Sentiment Analysis; Facial Expression Analysis; Question Generation; Natural Language Processing; BERT; Whisper STT