Interview Buddy: Refining language skills through personalized, automated interviews

Abstract

This paper presents Interview Buddy, a custom GenAI-powered chatbot designed to emulate oral proficiency interview formats such as the Oral Proficiency Interview (OPI) and Peace Corps Language Proficiency Interview (PCLPI). Developed as part of a three-month hybrid internship, Interview Buddy seeks to address the resource and scalability challenges inherent in personalized verbal language assessments by providing an accessible, automated training tool. The system integrates uploaded knowledge repositories, JSON-based behavioral instructions, and a phased interview ap- proach incorporating climbing and spiraling ques- tion difficulty. Quality assurance testing across multiple languages revealed both the potential of the system and several key limitations, including memory constraints, hallucinated content, and sequencing errors. To address these challenges, the paper explores potential solutions such as models with larger context windows, external memory management through a Flask microserver, and integration with workflow automation tools like n8n. While there is considerable room for refinement, Interview Buddy represents a promising step toward scalable, technology-driven solutions for language proficiency assessment.