SmartPath AI: An Integrated Full-Stack Framework for AI-Driven Government Exam Discovery, Eligibility Matching, and Adaptive Preparation

Abstract: Aspirants preparing for government recruitment examinations in India face a uniquely fragmented information landscape: notifications for national examinations such as the Union Public Service Commission (UPSC) Civil Services Examination, the Staff Selection Commission (SSC) Combined Graduate Level examination, Institute of Banking Personnel Selection (IBPS) recruitment, and Indian Railways recruitment, alongside state-level examinations conducted by the Karnataka Public Service Commission (KPSC), are published independently across dozens of authority-specific websites, each with its own format, eligibility clause, and application timeline. This paper presents SmartPath AI, a full- stack web platform that unifies exam discovery, eligibility screening, and personalized preparation for both national and Karnataka state government examinations behind a single interface. The system combines a scheduled web-scraping layer that continuously harvests notifications from official examination-authority websites with a human-moderated approval queue, a rule-based eligibility engine that evaluates a candidate's age, state of domicile, and academic qualification against each examination's criteria, and a course-to-career "wizard" that maps a candidate's educational background onto a ranked list of matching examinations and government job profiles. Layered on top of this discovery- and-matching foundation is a generative-AI content engine, built on the Anthropic Claude API, that produces exam- specific mock tests, topic-wise quizzes, structured study notes, and an adaptively recalibrated study timetable, together with a document-intelligence pipeline that extracts structured fields from scanned government notification PDFs and academic marksheets. A conversational AI study coach and a gamification layer comprising streaks, badges, a leaderboard, and peer study groups are added to sustain daily engagement. The platform is implemented with a React and Vite single-page frontend, a Python FastAPI backend backed by SQLAlchemy and SQLite, a lightweight Node.js/Express static-and-proxy layer, JSON-Web-Token-based authentication, and Razorpay-based subscription billing, and is deployable as an installable, offline-capable Progressive Web Application. Functional evaluation of the deployed system indicates that combining scraped-and-moderated official data, transparent rule-based eligibility logic, and generative AI content substantially reduces the manual effort an aspirant would otherwise spend locating, verifying, and preparing for relevant examinations.

Keywords: Government Exam Preparation, Eligibility Screening, Rule-Based Systems, Web Scraping, Generative Artificial Intelligence, Large Language Models, Conversational AI, Gamification, FastAPI, React.