Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{196037,
        author = {Shivam Lohar and Umar Farooq Azmi and Rakesh Choudhary and Moajez Lalani and Dr. Shikha Gupta},
        title = {Andrew: A Digital Platform for AI-Based Virtual Interview Assistant},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {1671-1680},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196037},
        abstract = {This study presents Andrew, an artificial intelligence- driven virtual assistant built to conduct domain-specific technical as- sessments across roles including Java Developer, Full-Stack Developer, Data Structures & Algorithms, DevOps Engineer, Frontend Developer, and Backend Developer. Moving beyond standard mock interview appli- cations, our platform measures candidate performance comprehensively by evaluating several distinct metrics, specifically answer quality, domain expertise, communication clarity, confidence levels, and problem-solving capabilities. The platform captures these metrics via computer vision- based proctoring (MediaPipe FaceMesh), audio analysis (faster-whisper speech recognition), alongside natural language processing capabilities (a fine-tuned Phi-3-mini utilizing QLoRA). We integrated a real-time behavioral integrity component to monitor gaze shifting, blink frequency, head positioning, and facial presence, allowing the software to calculate a proctoring score parallel to the main interview assessment. Following the session, the user receives an extensive feedback document featuring radar chart visualizations, a question-by-question performance breakdown, and detailed behavioral analytics. The entire architecture executes locally on standard consumer-grade hardware (RTX 2050, 4 GB VRAM) without any cloud dependencies during active sessions, making structured, data- informed interview practice widely accessible.},
        keywords = {AI, behavioral proctoring, computer vision, fine- tuning, natural language processing, QLoRA, speech recognition, virtual interview assistant.},
        month = {April},
        }