An Intelligent Academic Assistance Framework Using LLM Models

Abstract: AcademiAI is an intelligent academic assistance framework designed to support students and educators by unifying study-material comprehension, doubt-resolution, assignment evaluation, and personalized learning planning within a single system. It integrates document processing, Optical Character Recognition, Retrieval-Augmented Generation (RAG), and Generative AI-based conversational tutoring to help learners understand course content, get instant answers grounded in their own study material, and receive structured feedback on assignments and essays. AcademiAI accepts textbooks, lecture notes, syllabi, and handwritten class notes in PDF, DOCX, PPTX, or image form. Scanned and handwritten material is passed through Tesseract OCR to extract text, which is then chunked and embedded into a vector database for retrieval. Extracted content and user queries are routed through the Groq API to two large language models — a lightweight model for fast document summarization/embedding-support tasks and a larger instruction-tuned model (e.g., Llama 3.3 70B / GPT-OSS 120B) for conversational tutoring, essay feedback, and personalized study-plan generation. Core features include a Retrieval-Augmented Q&A engine grounded in the student's own materials, an AI tutor chatbot, an assignment/essay evaluation module, automatic quiz generation, a progress- tracking dashboard, secure login, and chat history. The combination of NLP, OCR, vector retrieval, and generative AI in AcademiAI demonstrates how multimodal, retrieval-grounded LLM systems can make academic support more personalized, scalable, and accessible.

Keywords: Generative AI, Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), Optical Character Recognition (OCR), Natural Language Processing (NLP), Vector Database, Academic Chatbot, Automated Essay Evaluation, Personalized Learning, Artificial Intelligence.