Your AI Study Companion — Powered by Gemma 4
Features • Architecture • Demo • Installation • Tech Stack • License
GemMate transforms how university students learn by combining Google's Gemma 4 E2B model with proven study science techniques. It's a cross-platform Flutter app that runs Gemma 4 100% locally — no cloud, no API keys, no data leaves your device.
💡 The Problem: Students struggle to create effective study materials from lectures and textbooks. Existing AI tools require cloud connectivity and raise privacy concerns.
✅ The Solution: GemMate runs Gemma 4 E2B on your own hardware, generating personalized flashcards, quizzes, and explanations — even on an airplane.
Chat with Gemma 4 E2B to understand complex concepts. Ask questions in any of 6 supported languages, and get bilingual explanations tailored to your level.
Generate flashcard decks from any chat conversation. Cards use the SM-2 spaced repetition algorithm for scientifically-optimized review schedules. Decks are displayed as beautiful fan-shaped card piles with flip animations.
| Front | Back |
 |
 |
AI-generated multiple-choice quizzes that test your understanding. Wrong answers automatically become flashcards for targeted review.
Photograph textbook pages, lecture slides, or handwritten notes. Gemma 4's vision capabilities extract and explain the content.
Tap the microphone to ask questions by voice — perfect for hands-free studying. Supports Chinese, English, Japanese, Korean, French, and Spanish.
Full UI localization and AI responses in: English, 简体中文, 日本語, 한국어, Français, Español.
Spaced repetition reminders, daily study prompts, and inactivity nudges keep you on track.
Beautiful neomorphic UI with dark/light mode, customizable accent colors, and adjustable font sizes.
GemMate uses a smart routing architecture that automatically selects the best available AI model:
┌─────────────────────────────────────────────────┐
│ 📱 PHONE │
│ GemMate Flutter App │
│ │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ Chat │ │ Cards │ │ Quiz │ │
│ └────┬─────┘ └────┬─────┘ └────┬─────┘ │
│ │ │ │ │
│ └──────────────┼──────────────┘ │
│ │ │
│ ┌───────▼────────┐ │
│ │ Smart Router │ │
│ └───┬───────┬─-──┘ │
│ │ │ │
│ ┌────────────▼─┐ ┌──▼───────────────┐ │
│ │ On-Device │ │ Ollama HTTP │ │
│ │ Gemma 4 E2B │ │ Connection │ │
│ │ (Offline) │ │ (WiFi LAN) │ │
│ └──────────────┘ └──────┬───────────┘ │
│ │ │
└──────────────────────────────┼──────────────────┘
│ WiFi (Local Network)
┌──────────────────────────────▼───────────────────┐
│ 💻 LAPTOP │
│ Ollama + Gemma 4 E4B │
│ (RTX 4060, <1s response) │
└──────────────────────────────────────────────────┘
| Condition | Model Used | Latency |
|---|---|---|
| WiFi + Laptop available | Gemma 4 E4B via Ollama (laptop GPU) | <1s |
| No WiFi, model installed | Gemma 4 E2B on-device (phone CPU) | 3-8s |
| WiFi + No laptop | Gemma 4 E2B on-device | 3-8s |
| No WiFi, no model | Prompt to download model | — |
📺 Watch the 3-minute demo video →
- Flutter 3.41+ (Install Flutter)
- Android device (Android 8.0+) or emulator
- For laptop AI: Ollama +
ollama pull gemma4:e2b
# Clone the repository
git clone https://github.com/linyeping/GemMate.git
cd GemMate
# Install dependencies
flutter pub get
# Run on connected device
flutter run
# Build APK
flutter build apk --release# Install Ollama (https://ollama.ai)
ollama pull gemma4:e2b
# Start with network access
OLLAMA_HOST=0.0.0.0:11434 ollama serve
# In GemMate Settings → Connection → Enter laptop IPOption A: Download in-app (Settings → Model Management → Download)
Option B: Manual install via ADB:
# Download from Hugging Face mirror (China)
curl -L -o gemma-4-E2B-it.litertlm "https://hf-mirror.com/litert-community/gemma-4-E2B-it-litert-lm/resolve/main/gemma-4-E2B-it.litertlm"
# Push to phone
adb push gemma-4-E2B-it.litertlm /sdcard/Download/
# In app: Settings → Model Management → Load from /sdcard/Download/| Component | Technology |
|---|---|
| AI Model | Gemma 4 E2B /Gemma 4 E4B |
| On-Device Runtime | LiteRT-LM via flutter_gemma |
| Local Server | Ollama (laptop, GPU-accelerated) |
| App Framework | Flutter 3.41 / Dart |
| Study Algorithm | SM-2 Spaced Repetition |
| Voice Input | speech_to_text |
| OCR / Vision | Gemma 4 multimodal (via Ollama) |
| Notifications | flutter_local_notifications |
| Storage | SharedPreferences + JSON |
| UI Design | Custom Neomorphic widgets |
lib/
├── main.dart # App entry point + model initialization
├── app/
│ ├── router.dart # Bottom navigation + page routing
│ └── theme.dart # Neumorphic theme (light/dark)
├── core/
│ ├── constants.dart # App constants + colors
│ ├── json_utils.dart # JSON parsing utilities
│ ├── text_utils.dart # Text processing and formatting
│ └── utils.dart # General helper functions
├── l10n/
│ ├── app_localizations.dart # i18n delegate
│ ├── locale_en.dart # English localization
│ ├── locale_es.dart # Spanish localization
│ ├── locale_fr.dart # French localization
│ ├── locale_ja.dart # Japanese localization
│ ├── locale_ko.dart # Korean localization
│ └── locale_zh.dart # Chinese localization
├── models/
│ ├── chat_message.dart # Chat message model
│ ├── chat_session.dart # Chat session model
│ ├── flashcard.dart # Flashcard with SM-2 fields + grouping
│ ├── quiz.dart # Quiz state model
│ ├── quiz_question.dart # Quiz question model
│ └── study_plan.dart # Spaced repetition schedule model
├── screens/
│ ├── capture_screen.dart # Camera / OCR feature
│ ├── chat_history_screen.dart # Chat session management
│ ├── chat_screen.dart # Main chat UI + voice input
│ ├── deck_study_screen.dart # Card flip study session
│ ├── flashcard_screen.dart # Deck gallery with fan piles
│ ├── home_screen.dart # Main dashboard
│ ├── onboarding_screen.dart # First-launch setup + model download
│ ├── paper_screen.dart # Document/exam paper analysis
│ ├── quiz_screen.dart # Interactive quiz UI
│ ├── review_screen.dart # Scheduled review dashboard
│ └── settings_screen.dart # Settings with sub-pages
├── services/
│ ├── flashcard_generator.dart # AI-powered flashcard creation
│ ├── local_gemma_service.dart # On-device Gemma 4 via flutter_gemma
│ ├── model_download_service.dart # Model download + mirror support
│ ├── notification_service.dart # Study reminders
│ ├── ollama_service.dart # HTTP client for Ollama API
│ ├── quiz_generator.dart # AI-powered quiz generation
│ ├── smart_router.dart # Smart model selection logic
│ ├── storage_service.dart # Local file/DB storage operations
│ └── study_tools.dart # Core study algorithms (SM-2, etc.)
├── stores/
│ ├── chat_store.dart # Chat session persistence
│ ├── connection_store.dart # Connection state management
│ ├── flashcard_store.dart # Flashcard persistence + groups
│ ├── locale_store.dart # Language preferences
│ └── theme_store.dart # Theme + font size preferences
└── widgets/
├── animated_avatar.dart # AI/User animated profile picture
├── chat_session_tile.dart # Chat history list item
├── color_scheme_picker.dart # Theme color selector
├── connection_indicator.dart # Connection status pill
├── download_progress_widget.dart # Model download status UI
├── flashcard_widget.dart # Individual flashcard UI
├── loading_indicator.dart # Custom loading animation
├── message_bubble.dart # Chat message bubble
├── model_badge.dart # Model source indicator (Edge/Laptop)
├── neumorphic_button.dart # Neumorphic button widget
├── neumorphic_container.dart # Neumorphic card widget
├── quick_action_chips.dart # Suggested prompt chips
└── quiz_option_tile.dart # Quiz multiple choice button
Sheng Wei — AI major at Gansu Political Science and Law University (GSUPL) in China. A Solo developer.
Previous Projects: InSeeVision (Gemma 3 accessibility project).
- GitHub: @linyeping
- Kaggle: linyeping
This project is licensed under the Apache License 2.0 — see the LICENSE file for details.
The Gemma 4 model is provided by Google under the Gemma Terms of Use.
Built with ❤️ for the Gemma 4 Good Hackathon 2026
Contact: yepinglin20@gmail.com | 201180946@qq.com
