mobile

2022-2023

Case Study

Revisable

AI-powered medical exam prep platform with 100K+ students, featuring Ruby AI trained on 1M+ medical documents.

Live Website

Key Metrics

100K+

Active Students

270K+

Flashcards

40K+

QBank Questions

4.5★

App Rating

Overview

Revisable is an AI-powered study platform I built as the sole developer for medical students preparing for competitive exams like NEET, USMLE, FMGE, and more. The platform serves 100,000+ active students with features including Ruby AI (trained on 1M+ medical documents), 270K+ flashcards with spaced repetition, and 40K+ QBank questions.

My Role

Solo Developer

I built the entire Revisable Flutter application from scratch - from UI/UX implementation to AI integration, payment processing, and performance optimization. The app achieved 100K+ downloads with a 4.5-star rating.

Challenges

Medical Content Accuracy

Medical education requires absolute accuracy. AI-generated explanations needed to be grounded in verified medical literature and guidelines, not hallucinated content.

Scale and Performance

Serving 100K+ students with 270K+ flashcards and real-time spaced repetition calculations required careful optimization of both client and server-side code.

Engagement and Retention

Medical students often burn out. The platform needed gamification, social features, and progress tracking to maintain engagement over months of study.

Multi-Exam Support

Different exams (NEET UG, NEET PG, USMLE, FMGE) have different syllabi and question patterns. The platform needed to serve all without fragmenting the codebase.

Solutions

Ruby AI with RAG

Built Ruby AI using Retrieval-Augmented Generation trained on 1M+ medical guidelines, textbooks, and research papers. This ensures explanations are grounded in verified medical literature.

Optimized Spaced Repetition

Implemented SM-2 algorithm with custom modifications for medical content. Cards are scheduled based on difficulty, topic relationships, and exam proximity.

Gamification System

Built comprehensive gamification including streaks, leaderboards, achievements, and study paths. These features significantly improved daily active user retention.

Modular Content Architecture

Designed a flexible content system where subjects, topics, and questions are tagged by exam type. Students see personalized content based on their target exam.

Technical Highlights

Ruby AI Integration

Integrated ChatGPT with custom prompting and RAG to create Ruby AI - a medical tutor that can explain concepts, answer doubts, and generate practice questions grounded in verified medical content.

PDF/Video to Flashcard Conversion

Built a feature that converts uploaded PDFs and video transcripts into structured flashcards in seconds using AI extraction and formatting.

Dual Payment Integration

Implemented both Razorpay (India) and Stripe (international) payment gateways with subscription management, trial periods, and revenue tracking.

Collaborative Flashcard Creation

Built a system where students can create and share flashcard decks, with voting and quality scoring to surface the best community content.

Performance Optimization

Achieved smooth 60fps scrolling through 270K+ flashcards using lazy loading, efficient state management, and careful widget tree optimization.

Results & Impact