Sankalp: AI-Powered Career Guidance System

Abstract

Traditional career guidance systems often lack personalization, adaptability, and accessibility, leading to suboptimal decisions in a rapidly changing job market. This study introduces Sankalp, an AI-driven career guidance framework that combines semantic reasoning, emotion analysis, and adaptive learning to provide inclusive and explainable recommendations. Built on a multi-agent hybrid architecture, Sankalp integrates modules for semantic matching using Sentence-BERT (SBERT), emotion-aware feed- back through Valence Aware Dictionary and Sentiment Reasoner (VADER) sentiment analysis, a Naukri API–based job trends agent, a knowledge graph, and a voice assistant that supports English, Hindi, and Kannada. A reinforcement learning–based fusion layer dynamically refines recommendations based on real-time user feedback, modeling career choice as a life-span adaptive process. It was evaluated against a Curated Ground Truth (CGT) dataset using 160 participant profiles, Sankalp demonstrated superior performance, achieving a Top-1 Hit Rate of 77.5%, a Top-3 Hit Rate of 90.5%, and a high mean user satisfaction score of 4.71/5.0. The latency analysis confirmed high operational efficiency, with an average end-to-end response time of 1.25 s. Through its inclusive design and adaptive, explainable hybrid AI, Sankalp enhances the accessibility, reliability, and ethical delivery of career guidance services, aligning with the global sustainable development goals.

Ähnliche Arbeiten