<S>aransh-OS
booting...
PROJECT

PrepStack AI

Next.jsGemini 2.5BullMQ & RedisTailwind CSS v4
PrepStack AI Platform
Year2026
engagement

PrepStack AI is a high-fidelity, premium SaaS platform designed for educators and institutions to instantly formulate curriculum-aligned question papers and solution schemes. By parsing raw course materials (PDFs, Microsoft Word .docx files, and images) and leveraging Gemini 2.5 Flash, PrepStack AI generates beautiful, printable A4 question sheets complete with institution-branded letterheads, custom visual badges, and separated teacher answer keys.

🌟 Key Engineering Highlights

  • Generative Question Formulation: Calibrates dynamic prompts using Gemini 2.5 Flash to generate structured multi-section question sheets parameterized by customizable Marks, Topic, Class Grade, Question Types, and specific teacher instructions.
  • Dual-Column Authentication & Interactive Onboarding: Features password visibility micro-interactions and a full-bleed visual grid showcasing branding graphics, alongside a 3-step onboarding wizard for educators.
  • Strict Multi-User Session Isolation: Dynamically scopes assignments and profile parameters using account-specific tokens (PrepStackai_profile_${userEmail} and x-user-email headers), allowing multiple teachers to test in the same browser sandbox without data leakage.
  • Multi-Format Document Parsing: Employs magic-byte binary header matching and the Mammoth parser to dynamically extract curriculum text from uploaded PDFs, images, and .docx files.
  • Resilient Queued Generation: Implements a background task pipeline using BullMQ and Redis, backed by an exponential backoff wrapper (generateWithRetry) to handle rate limits and transient upstream outages smoothly.
  • Premium Printable A4 PDF Exports: Independent high-definition PDF compilation for Question Papers and Answer Keys using jspdf and html2canvas-pro with Tailwind CSS v4 wide-gamut color parsing (oklab(), oklch()) and zero-student leak print overlays.
services
Full-Stack-DevelopmentAI-EngineeringPDF-Compilation EngineQueue-Architecture

Up Next

Nice things people say about my work

collaborations
MyPic
👋

Let’s work together on your next project

let's-work-together →let's-work-together →