LLM-RAG-Based Student Progression Tracking Pl...

LLM-RAG-Based Student Progression Tracking Platform for Academic and Career Guidance

Author Name : Sudhaveni Arun, Bandari Manasa, Aemuganti Abheesta, Badam Prashanth

DOI: https://doi.org/10.56025/IJARESM.141026205

 

ABSTRACT

Higher education institutions face a widening gap between academic curricula and the skills that employers expect, yet conventional campus Enterprise Resource Planning (ERP) systems only record grades and attendance; they do not diagnose weaknesses or guide preparation. This paper presents AI Campus OS, a full-stack student progression tracking platform that unifies academic monitoring, early risk detection, AI tutoring, and placement preparation for five user roles—students, faculty, parents, placement officers, and administrators—under role-based access control. The system is built on an asynchronous FastAPI backend, a React 18 and TypeScript client, and a PostgreSQL 16 data layer, and it reaches large language and speech models through NVIDIA NIM inference microservices. Every LLM request uses retrieval-augmented generation (RAG): a context builder retrieves the student's own academic, coding, and résumé records from the institutional database and places them in the prompt, so that tutoring, feedback, and career guidance are grounded in that student's data. Its AI services include a speech-enabled mock interview engine that transcribes spoken answers with NVIDIA Canary or Whisper Large v3 Turbo and scores them on a five-dimension rubric; a Socratic tutor with two levels of explanation depth, automatic quiz generation, and day-by-day study plans; and an ATS résumé reviewer for PDF, DOCX, and TXT files. Live coding activity is collected from LeetCode and GitHub.