Kaushik Bhingradiya's Work | ContraWork by Kaushik BhingradiyaKaushik Bhingradiya
Mobile MVP Developer for Startups | Flutter, React Native.
VEDV
One app, four roles, turning printed textbooks into a connected school platform.
Overview
VEDV is a Flutter app (iOS, Android, web/desktop) that serves as VEDV Publication's all-in-one school platform. From a single codebase it presents four different experiences based on login , Student, Teacher, School Admin, or Guest/Parent with UI gated by role flags. At its core, it turns printed textbooks into connected digital lessons: students scan a QR code on a physical book and land directly on the matching video, PDF, quiz, or test. Around that sits a full learning library, three Gemini-powered AI assistants, and a School Hub that auto-generates school documents all layered over one educational-safety policy that keeps AI responses age- and subject-appropriate.
For teachers and admins, the same app doubles as a school-operations backbone: attendance, marks, marksheet and ID-card generation, certificates, fees, notices, calendars, and real-time 1:1 chat.
The Problem
Schools run on disconnected tools one system for content, another for attendance, spreadsheets for fees, WhatsApp for parent updates with no link between a printed textbook and any digital follow-up. A publisher's book is a dead end once printed. The result is duplicated data entry and no single, trustworthy view into a child's attendance, results, or fees. VEDV bridges that gap: one role-aware app where the QR scan turns a static book into an interactive lesson, and the same platform runs the school's day-to-day operations so content and operations reinforce each other instead of living in silos.
Key Features
QR Book Scan — Scans a printed textbook's QR code and opens the exact matching content (video, PDF, quiz, audiobook), filtered by role, students see learning material, teachers additionally get lesson plans and answer keys.
AI Assistant with Chapter-PDF & Multi-Format Export (Unique) — Type, speak, or photograph a question; the AI answers grounded in the actual chapter PDF and can export a real downloadable worksheet, lesson, or presentation as PDF, PowerPoint, or Word.
School Hub AI Design Generator (Unique) — Describes a poster, ID card, or notice in plain language, asks smart follow-up questions, then generates a print-ready design using the school's real logo and brand colors.
STEAM-L AI Tutoring Suite (Unique) — Six subjects (Science, Tech, Engineering, Arts, Math, Languages), each with an AI tutor, auto-generated quizzes, an English corrector, and a full voice-mode conversation gamified with streaks, XP, and hearts.
Tech Stack
Mobile: Flutter · Dart · GetX (state, routing, DI) · GetStorage (local persistence).
Backend: REST API via Dio · Laravel · MySQL · Firebase (Auth, Realtime Database for chat/presence, Cloud Messaging).
AI: Google Gemini (routed through a shared safety-guardrail layer).
Infra: GitHub · Sentry (crash/performance monitoring).
Results & Key Takeaways
AI responses are unreliable in raw form the real fix was building a robust clean-up layer that recovers useful content even from malformed AI output, rather than trusting the model directly.
Never hard-wire a single AI model each request tries a preferred model and falls back through alternatives with retries, since models get updated or retired often.
Security gap flagged for next iteration: AI API keys currently live inside the app itself and should move server-side before wider distribution.
The same AI-handling logic (retries, fallbacks, safety rules) was duplicated across features next step is consolidating it into one shared component.
The "one app, four roles" model worked well but role-permission logic is scattered across screens recommend centralizing role management into a single control point. CarHeat: Vehicle-Specific Parked-Car Heat Alerts Activity: City-Based Activity Discovery & Booking Platform
Activity is a Kuwait-based mobile platform built to make discovering and booking local experiences simpler. Instead of searching separately for movies, parties, events, games, workshops, kids’ activities, fitness sessions, food experiences, and other things to do, users can discover activities around them, choose a time slot, complete the booking, and manage everything from one app.
The platform connects users with local vendors who organize activities around the city, turning activity discovery into a complete journey from finding something interesting to showing up for the experience.
The problem
Finding something to do often means checking multiple platforms, social pages, event listings, and individual booking services.
The challenge was to create a single experience where users could discover different types of activities based on their location and interests, understand the available options, select a suitable time, make a payment, and keep track of their bookings without switching between different services.
Activity brings those steps together by combining location-based discovery, search, filtering, time-slot booking, payments, favorites, and booking management in one mobile experience.
My role
Flutter development for iOS and Android
UI/UX implementation for activity discovery and booking
Location-based activity discovery and map integration
Search, filtering, favorites, and activity browsing
Booking and time-slot selection flows
Payment and booking confirmation integration
QR-based booking and attendance experience
Quality assurance and release testing.
Key features
Location-based activity discovery
Activity is built around discovering things to do nearby. Users can browse activities based on their location and use the interactive map to visualize available experiences around them. Activities can be filtered by location, type, and time slot.
Activity discovery and categorization
The app brings different types of experiences into one place, including movie screenings, parties, social events, indoor and outdoor games, kids’ activities, workshops, fitness, yoga, wellness, art, cooking, food experiences, and more.
Search and advanced filters
With a large and varied activity catalogue, simply displaying activities in a list would make discovery harder.
Activity details and time-slot booking
Once users find an activity, they can open its details, review the available information, select a preferred date, choose a suitable time slot, and book their spot.
The supplied booking flow shows this progression directly: activity details → date selection → time selection → Book Now.
Technical approach
Mobile: Flutter, Dart
Platforms: iOS and Android
Core modules: Location services, map-based discovery, search and filters, activity catalogue, favourites, booking management, time-slot selection, payments, QR-based booking confirmation
Integrations: Maps/location services, payment gateways, backend APIs
Product areas: User discovery, activity/vendor listings, scheduling, booking, payment, and attendance
The application was designed as a complete mobile booking experience rather than simply an activity catalogue, with the major product flows connected from discovery through attendance.
Engineering and product decisions
The biggest product challenge was balancing a large variety of activities with a simple discovery experience.
A user looking for a specific activity should be able to search and filter directly, while someone who simply wants to find something interesting should be able to browse suggested, featured, and recently viewed activities.
The booking flow was therefore kept closely connected to the activity itself: users move from the activity card to its details, select a date and available time, complete payment, and then access the reservation through My Bookings.
The project's clearest product distinction is this end-to-end flow: discover something to do → choose when to go → book the spot → manage the reservation → show up with the booking confirmation.
Outcome
Activity shipped as a mobile activity discovery and booking application for the Kuwait market, connecting users with local vendors and experiences across entertainment, activities, workshops, wellness, food, and other categories. The Android application is publicly available on Google Play and was updated on April 5, 2026.
The result is a single mobile experience that brings together local discovery, map exploration, search, scheduling, payments, favourites, booking management, and QR-based attendance.
The product idea is simple:
Find an activity.
Book your spot.
Just show up. Overview
OneBrain is a mobile AI super-app by DIGITX that ends the habit of switching between ChatGPT, Gemini, Claude and a pile of single-purpose generators. You sign in once with Google or email, and the same app gives you a model-switching chat and a creation studio for images, video, voice and text utilities. The studio is built for generations that take a while: you can start a render, leave the app, and come back to a finished result.
Key Features
Multi-Model Chat (Unique): ChatGPT, Claude, Gemini, DeepSeek, Grok and Leo sit in one picker. Switch models mid-conversation without leaving the thread.
Streaming Responses: Replies arrive token by token, and each one can be copied, read aloud or downloaded.
Voice Input: Type or speak, with language selection built into the message bar.
Imagine (Image Generation): Six image models, with controls for aspect ratio, model and quality. Results land in a filterable gallery of images and videos.
Video Generation (Unique): Eight video models in one grid: VEO, Runway, Hailuo 2 Fast, Seedance, Lightricks and Wan, among others. Premium models like VEO are gated by media tokens, with an upgrade prompt shown in context.
Audio & Voice Studio: ElevenLabs-powered text-to-speech, speech-to-speech and voice cloning, with a searchable voice library split into Popular, My Voices and Cloned.
AI Humanizer: A standalone tool that rewrites AI-sounding text so it reads like a person wrote it.
AI Memory & Usage Controls: Personal memory settings, usage tracking, integrations and appearance options, all in one settings area.
Persistent Library: Every generation is saved in one place.
Tech Stack
Flutter · Dart · BLoC · Firebase · Dio · Async Generation Pipeline · Persistent Job Queue · ElevenLabs · Google Sign-In
Results
Shipped to the Play Store as a working multi-model AI app: six chat models, 14 image and video generators, voice cloning and a text humanizer, all in one native mobile experience.