A RAG based Personal Placement Assistant System using Large Language Models for Customized Interview Preparation

This paper introduces a Personal Placement Assistant (PPA) framework that utilizes advanced Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) to automate and personalize job placement preparation. The system integrates Natural Language Processing (NLP) techniques, including text embedding using the all-MiniLM-L6-v2 transformer model and semantic retrieval using ChromaDB for accurate resume analysis and context-aware question generation. The PPA is structured into three core components: the Retriever, using PyMuPDF for resume parsing and recursive text chunking for efficient vector storage and search; the Analyzer, employing the Google Gemini-1.5-flash model for domain extraction and percentage-based content profiling; and the Generator, which produces domain-specific MCQs, coding challenges, and interview questions aligned with Bloom’s Taxonomy. RAG enhances the system’s ability to integrate external knowledge, improving the contextual relevance of the generated content. Evaluation results demonstrate an 83.77% accuracy in domainspecific extraction and question generation, confirming the PPA’s effectiveness in automating personalized job preparation across industries.