A full-stack exam intelligence platform for searching previous-year questions, building practice papers, controlling premium access, and supporting public, student, and admin workflows.
Building an exam-intelligence platform where search, practice, entitlement, and publishing share one reliable product flow.
PYQPath is a multi-exam question intelligence and practice platform. It started with UPSC Mains previous-year questions and grew into a product with public discovery, registered student workflows, premium question access, paper building, saved papers, downloadable answer-writing layouts, and a dedicated admin console.
Previous-year questions are valuable only when students can retrieve them by the way they actually study: exam, paper, subject, topic, year, marks, keywords, and combinations of those dimensions. The engineering problem was not just building a searchable list; it was creating a consistent content model, protecting premium workflows at the API layer, generating useful practice artifacts, and still keeping public pages discoverable.
I worked across the full product path: React/Vite application flows, Express APIs, JWT-based authentication, MySQL schema and migrations, question import and search-quality tooling, entitlement enforcement, paper-building workflows, admin operations, SEO generation and public-route prerendering, plus deployment-oriented release checks for Hostinger.
React routes separate public discovery from authenticated student and admin experiences. Public pages remain lightweight while advanced question and paper workflows live behind account and entitlement checks.
React · Vite · React Router · HelmetExpress exposes question, paper, account, preview, and administrative endpoints with authentication, rate limiting, validation, and server-side access enforcement.
Node.js · Express 5 · JWT · Zod · HelmetMySQL stores the canonical question bank and the dimensions needed for filtering, papers, access state, and content administration. Migrations evolve the product without replacing the data foundation.
MySQL · mysql2 · SQL migrationsImport, verification, reindexing, search-quality tests, paper creation, saved papers, and answer-writing output turn the question bank into repeatable study workflows.
Search scripts · Paper builder · jsPDF · html2canvasPublic preview endpoints return only explicitly sampled questions while paid APIs enforce entitlement on the server. Build scripts generate SEO assets and prerender public routes for discoverability.
Entitlements · Preview API · Prerender · HostingerHiding frontend links is not security. Complete question and paper endpoints require authentication and entitlement, while public visitors use separate preview endpoints that only return rows explicitly marked as samples.
Search quality depends on structured exam, stage, paper, subject, topic, year, marks, and related metadata. Keeping that foundation canonical makes filtering and future intelligence features easier to evolve.
Public routes get SEO generation and prerendering, while authenticated student/admin workflows remain application routes. Search engines receive useful public content without exposing premium data.
The repository includes focused scripts for search quality, scope contracts, taxonomy, public preview, payments, SEO, and pre-launch checks so product changes can be verified against intended behavior.
A question can participate in several study dimensions. The data model and import/verification tooling had to keep those dimensions coherent enough for combined search and paper generation.
The public preview was separated from the full question library at both frontend and backend levels, with dedicated preview APIs and explicit sample flags.
Public marketing and content routes need crawlable output while the student product behaves like an application. Build-time SEO generation and prerendering bridge those different requirements.
As paper building, saved papers, access control, admin tooling, payments, writing workflows, and release checks were added, the architecture had to keep the question bank as the stable core instead of fragmenting into separate features.
The PYQPath production repository is private. This case study describes implemented product behavior and repository structure without exposing private data, credentials, or proprietary question content.
← BACK TO SELECTED WORK