SMART SCHEME & SCHOLARSHIP RECOMMENDATION PORTAL

The Smart Scheme & Scholarship Recommendation Portal shows one workable way to help citizens find and prepare for government benefits while remaining trustworthy. Its core is a deterministic, explainable eligibility engine that uses three-valued logic, so that a missing fact becomes an explicit request for information instead of a misleading verdict. Around this core, the system provides a verified, versioned scheme catalogue, a structured and consent-based citizen profile, a secure document vault with a mandatory confirmation step, scheduled deadline and expiry notifications, a grounded AI assistant that may explain but never decide, and administrative analytics with an append-only audit trail. The architecture is a modular monolith in React, TypeScript, Express, Prisma and PostgreSQL, organised into ten modules with a deliberate separation between ground truth and assistance. This separation allows assistive components such as the AI provider or OCR engine to fail without affecting discovery or eligibility. Security and operational concerns are treated as design requirements, with short-lived and rotating tokens, role-based access, validated input, hardened headers, rate limiting, redacted logging, containerised deployment, backups and a documented recovery plan. The repository records 210 automated tests and a zero-warning lint policy. The work also has clear boundaries. The document extraction provider is currently a heuristic stand-in, rules have not been validated by awarding authorities, and no user study, load test or penetration test has been performed. The portal’s guidance is indicative, and final decisions remain with the responsible departments and institutions. Within these limits, the project shows that explainable rule evaluation, careful handling of missing information and restricted use of AI can be combined into a platform that is both useful and honest about what it does not know

[1] S. C. Kleene. Introduction to Metamathematics. North-Holland, Amsterdam, 1952. [2] F. Ricci, L. Rokach and B. Shapira (eds.). Recommender Systems Handbook, 2nd ed. Springer, 2015. [3] P. Lewis et al. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Advances in Neural Information Processing Systems, 33, 2020, pp. 9459-9474. https://arxiv.org/abs/2005.11401 [4] Tesseract OCR contributors. Tesseract User Manual. Official documentation. https://tesseract-ocr.github.io/tessdoc/ [5] OWASP Foundation. OWASP Top 10 Web Application Security Risks. https://owasp.org/www-project-top-ten/ [6] M. Jones, J. Bradley and N. Sakimura. JSON Web Token (JWT). RFC 7519, IETF, 2015. https://www.rfc-editor.org/rfc/rfc7519 [7] N. Provos and D. Mazieres. A Future-Adaptable Password Scheme. Proceedings of the USENIX Annual Technical Conference, 1999. [8] Government of India. The Digital Personal Data Protection Act, 2023. Ministry of Electronics and Information Technology. [9] Meta Open Source. React Documentation. https://react.dev [10] Government of India. National Scholarship Portal. https://scholarships.gov.in