Requests for Packaging: Vanderwaals - AI-Powered Wallpaper App with On-Device Machine Learning

Title:

Vanderwaals: Learns You - AI-Powered Wallpaper App with On-Device Machine Learning

Description:

Application Details

Field Value
App Name Vanderwaals: Learns You
Package ID me.avinas.vanderwaals
Version 2.7.0
Version Code 270
License AGPL-3.0-or-later
Category Personalization / Graphics
Minimum Android 12.0 (SDK 31)
Target Android 15 (SDK 36)
Developer Confused Coconut
Email hi@avinas.me

Description

Vanderwaals is an intelligent Android wallpaper application that uses on-device machine learning to understand your aesthetic preferences and automatically deliver fresh wallpapers you'll love.

Key Features

  • 🤖 On-device AI personalization
  • 📱 Two modes: Auto Mode (discover wallpapers) & Personalize Mode (upload favorites)
  • ⚡ Smart auto-change triggers: Unlock, hourly, daily, or manual
  • 🎨 Material 3 design with dynamic colors (Android 12+)
  • 🔒 100% offline - zero cloud tracking, zero analytics
  • 📚 Thousands of curated wallpapers from GitHub & Bing
  • 💾 History/Timeline view with quick actions
  • 🧠 Adaptive learning algorithm (EMA-based)
  • 🎯 Smart crop with saliency detection
  • 🌙 Dark theme support

Why F-Droid?

This app is a perfect fit for F-Droid:

✅ Open Source - AGPL-3.0-or-later license
✅ Privacy-First - 100% offline AI, zero analytics, zero tracking
✅ No Tracking SDKs - No Firebase, Google Analytics, or proprietary tracking
✅ No Ads - Completely ad-free with zero ad networks
✅ Transparent - Full source code available on GitHub, fully audit-able
✅ Reproducible Builds - Builds are reproducible from source
✅ Proper Permissions - All 9 permissions justified and necessary
✅ Modern Design - Material Design 3 compliant
✅ No Anti-Features - Zero issues with F-Droid policies
✅ Actively Maintained - Regular updates and bug fixes

Technical Information

Language: Kotlin
Framework: Jetpack Compose
Build System: Gradle (Android Gradle Plugin)
Signing: SHA-256 fingerprint configured

Core Dependencies (All Open-Source)

  • TensorFlow Lite (on-device ML inference)
  • Jetpack Compose (modern UI framework)
  • Kotlin Coroutines (async operations)
  • Hilt (dependency injection)
  • Room Database (local data persistence)
  • Material 3 Components
  • Jetpack Libraries (lifecycle, navigation, etc.)

Machine Learning Details

  • Model: MobileNetV3 (quantized, on-device)
  • Learning: Exponential Moving Average (EMA) algorithm
  • Feedback: Implicit learning from wallpaper display duration
  • Privacy: 100% local computation, zero cloud sync

Permissions Justified

Permission Purpose Justification
INTERNET Download wallpapers & sync catalog Essential for fetching wallpaper library
SET_WALLPAPER Apply wallpapers to device Core app functionality
READ_MEDIA_IMAGES Access user's gallery for personalization uploads User selects images to help train preferences
ACCESS_NETWORK_STATE Check connectivity before downloading Avoid errors on network disconnect
POST_NOTIFICATIONS Send wallpaper change notifications User-facing feature for auto-change alerts
FOREGROUND_SERVICE Background wallpaper sync & changes Required for scheduled operations
FOREGROUND_SERVICE_DATA_SYNC Sync wallpaper catalog Background data sync for library updates
REQUEST_IGNORE_BATTERY_OPTIMIZATIONS Maintain auto-change schedule Ensures scheduled wallpapers work reliably
RECEIVE_BOOT_COMPLETED Resume automation after device restart Persistent automation across reboots
SCHEDULE_EXACT_ALARM Precise daily change scheduling For user-selected exact time triggers

Source Code & Links

Link URL
GitHub Repository https://github.com/avinaxhroy/Vanderwaals
Source Code https://github.com/avinaxhroy/Vanderwaals (Full source, AGPL-3.0)
Issue Tracker https://github.com/avinaxhroy/Vanderwaals/issues
Releases https://github.com/avinaxhroy/Vanderwaals/releases
Latest Release v2.7.0 (November 20, 2025)

Build Information

Build Status: ✅ Ready to build from source
Reproducible: ✅ Yes
NDK Version: r25b

How to Build

git clone https://github.com/avinaxhroy/Vanderwaals.git
cd Vanderwaals
./gradlew clean assembleRelease

The APK will be generated in: app/build/outputs/apk/release/

Privacy & Security

✅ Zero Data Collection - No personal data collection
✅ Zero Cloud Sync - All processing local only
✅ Zero Analytics - No analytics libraries or trackers
✅ Zero Ads - No ad networks or SDKs
✅ Offline AI - Machine learning happens entirely on-device
✅ Audit-able - Full source code for security review

User preferences are stored locally on the device and never transmitted or tracked.

Testing

  • ✅ Tested on Android 12, 13, 14, 15
  • ✅ All features working correctly
  • ✅ No crashes or runtime errors
  • ✅ Permissions functioning as expected
  • ✅ Material 3 design verified
  • ✅ Dynamic colors tested
  • ✅ Dark theme verified
  • ✅ Auto-change automation tested

Screenshots

7 high-quality app screenshots available:

  1. Auto Mode overview & wallpaper discovery
  2. Personalize Mode with gallery upload
  3. History/Timeline view
  4. Settings & automation configuration
  5. Material 3 design showcase
  6. Dynamic colors (Android 12+)
  7. Dark theme support

App icon: 140 KB PNG format

Metadata

Complete F-Droid metadata provided in source repository:

  • ✅ App title and descriptions (en-US)
  • ✅ 7 high-quality screenshots
  • ✅ 140 KB app icon
  • ✅ Feature list and descriptions
  • ✅ Build instructions
  • ✅ License information
  • ✅ All required metadata fields

Location: metadata/en-US/ and .fdroid/metadata/me.avinas.vanderwaals.yml

Additional Information

Why This App Matters

Vanderwaals represents a unique approach to wallpaper personalization by bringing cutting-edge on-device machine learning to Android in a privacy-respecting way. It's the only wallpaper app with true on-device ML learning capabilities - all competitors either use cloud-based AI or simple algorithms.

Unique Selling Points

  1. True On-Device ML - No cloud, no tracking
  2. Intelligent Personalization - Learns what you actually like
  3. Completely Open Source - AGPL-3.0-or-later
  4. Privacy-First Design - Zero analytics by default
  5. Beautiful UI - Modern Material 3 design with dynamic colors
  6. Well-Maintained - Active development and regular updates
  7. Properly Documented - Clear code with inline documentation
  8. Production Ready - v2.7.0 stable release

Code Quality

  • ✅ Follows Kotlin best practices and conventions
  • ✅ Well-organized package structure
  • ✅ Clear separation of concerns (UI, ML, Data)
  • ✅ Comprehensive error handling
  • ✅ Proper resource management
  • ✅ Accessibility support
  • ✅ Regular updates and maintenance

Checklist

  • ✅ App complies with F-Droid inclusion criteria
  • ✅ Source code is openly available on GitHub
  • ✅ App is not already listed in F-Droid
  • ✅ App has not been previously requested
  • ✅ Developer has been notified (I am the developer)
  • ✅ No proprietary or tracking dependencies
  • ✅ License is compatible with F-Droid (AGPL-3.0)
  • ✅ All metadata is complete and accurate
  • ✅ Screenshots and icon provided
  • ✅ Build configuration is functional