Abstract

The increasing volume and diversity of student performance data have exposed significant limitations in traditional training and placement systems, which primarily rely on static eligibility criteria, manual shortlisting processes, and delayed communication mechanisms. These systems also lack the capability to leverage predictive analytics, resulting in minimal personalized, data-driven insights to enhance student employability based on individual skills and qualifications. To address these challenges, this study proposes an AI-driven education and placement platform that integrates machine learning-based placement prediction with conversational assistance and intelligent job matching. The system utilizes XGBoost for predictive modeling, Sentence-BERT embeddings for semantic skill representation, SHAP for explainable insights, and Retrieval-Augmented Generation (RAG)-based chatbots to provide real-time guidance and interview preparation support. The platform is implemented using FastAPI and deployed on cloud infrastructure, with automated email notification systems enabling real-time user interaction. The proposed system was evaluated using a dataset of 1,200 student records, incorporating academic, skill-based, and experiential attributes. Experimental results demonstrate an accuracy ranging from 88% to 90%, along with strong performance across multiple evaluation metrics, including precision, recall, F1-score, and ROC-AUC. Additionally, the system achieved low inference latency (<150 ms) and maintained stable performance under concurrent usage conditions.

Graphical Abstract

Novelty Statement

An integrated AI-driven placement platform combining predictive analytics and conversational assistance to enable real-time, scalable, and data-driven decision-making in institutional training and placement systems.