Exmentify

AI Web Application. An exam-agnostic AI study companion — built on a GATE CS foundation, designed to scale across every major Postgraduate competitive exam.

Comment

Inspiration

Competitive exam prep in India is fragmented — students juggle multiple apps, scattered PDFs, and YouTube playlists just to prepare for a single exam like GATE CS. As a student myself, I felt this pain firsthand: every new exam meant starting over with a new tool, a new interface, and a new set of resources. I wanted to build one AI-powered companion that adapts to whichever exam a student is preparing for, instead of forcing them to relearn a new platform every time.

What it does

Exmentify is an exam-agnostic AI study companion. For its MVP, it's fully built out for the full GATE CS syllabus, offering:

  • AI-generated explanations and practice questions tailored to the GATE CS syllabus
  • An AI chat assistant that answers doubts in context, instead of sending students off to search elsewhere
  • A roadmap to expand into UGC NET CS, Campus Placement prep, CAT, and UPSC CSE — all marked "Coming Soon" in-app, so the platform can scale to any competitive exam without a redesign

How I built it

  • Backend: Python 3.11 with FastAPI, handling the core application logic and API layer
  • AI layer: Google's Gemini API, powering explanations, question generation, and adaptive responses
  • Infrastructure: Dockerized the app for a consistent, portable build, and deployed it on AWS Elastic Beanstalk with CloudFront in front for HTTPS, giving us scalable, production-style hosting
  • I started from a full PRD before writing any code, which forced me to scope the MVP tightly around the full GATE CS syllabus while designing the data model to generalize across future exams from day one

Challenges I ran into

  • Getting Gemini to generate explanations that stay accurate and syllabus-aligned, rather than generic textbook answers, took careful prompt design and iteration
  • Structuring the data model so it's genuinely exam-agnostic — not hardcoded to GATE CS — while still shipping a focused MVP on time
  • Working within free-tier constraints (Gemini API quotas, AWS account limitations) forced real architectural decisions rather than "throw more resources at it" solutions

Accomplishments that I am proud of

  • Shipping a working full-GATE-CS-syllabus MVP end-to-end within the hackathon timeline
  • Designing the architecture so adding a new exam is a configuration/content change, not a rewrite

What I learned

  • How to scope an ambitious, multi-exam vision down into a shippable MVP without losing the bigger architecture
  • Prompt engineering matters as much as backend architecture when building an AI-first product

What's next for Exmentify

  • Rolling out UGC NET CS, Campus Placement, CAT, and UPSC CSE modules
  • Expanding the AI layer with personalized, adaptive study plans that use performance on practice questions to resurface a student's weak topics

Built With

Share this project:

Updates

Submission history