---

MoveYourBody

MoveYourBody is a privacy-first, on-device fitness application designed to help you stay consistent with your goals. It provides personalized, short micro-workout sessions (5–7 minutes) that adapt based on your feedback and health conditions. By running completely offline, MoveYourBody combines rule-based filtering and lightweight semantic matching to ensure your workout data stays entirely private while delivering safe, relevant, and engaging exercises.

---

🚀 Features

• Micro-Workout Sessions: Stay consistent with short, 5-7 minute micro-sessions. You'll receive 2-3 sessions per day, with each session consisting of 3 exercises, complete with built-in notifications.
• Personalized Exercise Selection: Uses an on-device embedding pipeline to match exercises to your goals. It avoids recently performed exercises to prevent boredom, filters out unsafe movements based on your injuries, and adapts difficulty through post-workout feedback.
• High-Quality Animations & Guidance: Understand every movement with crystal-clear animations and instructions that teach you the correct posture.
• Voice-Controlled Interface: Enjoy a hands-free workout experience with voice commands to start, pause, skip, or repeat instructions during a session.
• Body Focus Workouts: Want to target a specific muscle group? Choose from pre-defined sessions tailored for quick, muscle-specific training.
• Custom Workout Creation: Take full control. Explore a database of over 100+ exercises to learn, mix, and build your own custom, flexible workout routines.
• Progress & Motivation: Visualize your consistency and fitness journey with an interactive stats screen, calendar views, and performance charts.
• Privacy-First & 100% Offline: Everything runs on-device. Your health data, feedback, and embedded semantic searches never leave your phone.

---

💻 Tech Stack

Mobile Frontend

• Framework: Flutter
• State Management: Riverpod
• Routing: GoRouter
• UI/Animations: Lottie, Video Player, Google Fonts, FL Chart

Local Backend & Data

• Database: SQLite (sqflite)
• Caching: Shared Preferences, Flutter Cache Manager
• Search Utilities: Fuzzy string matching

On-Device AI

• Inference: Flutter ONNX Runtime
• Tokenization: Dart WordPiece
• Models: all-MiniLM-L6-v2 (Quantized ONNX model running directly on-device via assets/models/model_quantized.onnx)

---

✅ Project Checklist

• The mobile app:
  • has an About page containing the Stability Nexus's logo and pointing to the social media accounts of the Stability Nexus.
  • is available for download as a release in this repo.
  • is available in the relevant app stores.
• The AI/ML components:
  • LLM/model selection and configuration are documented.
    • Model Selection: all-MiniLM-L6-v2 was chosen because it is designed specifically for sentence similarity tasks. It provides low latency and fast inference which is highly suitable for on-device cases (see Research Paper).
    • Configuration: ONNX models are quantized and bundled locally in assets, with tokenization logic strictly handled on-device.
    • 
  • Prompts and system instructions are version-controlled.
  • Content safety and moderation mechanisms are implemented.
  • API keys and rate limits are properly managed.

---

🔗 Repository Links

1. Main Repository

---

🏗️ Architecture

1. High-Level MVVM Architecture

The application follows a robust Model-View-ViewModel (MVVM) with Repository Pattern architecture, utilizing Riverpod as the reactive state-management (ViewModel) layer to keep the UI strictly separated from the business and data logic.

graph TD
    %% Define Styles
    classDef ui fill:#1A2E1A,stroke:#4CAF50,stroke-width:2px,color:#fff;
    classDef vm fill:#254025,stroke:#81C784,stroke-width:2px,color:#fff;
    classDef model fill:#122412,stroke:#2E7D32,stroke-width:2px,color:#fff;
    classDef data fill:#0A1A0A,stroke:#66BB6A,stroke-width:2px,color:#fff;

    subgraph View Layer [View Layer]
        UI[Flutter UI / Screens]:::ui
        Widgets[Custom Widgets]:::ui
    end

    subgraph ViewModel Layer [ViewModel Layer]
        Riverpod[Riverpod Notifiers & Providers]:::vm
        Router[GoRouter State]:::vm
    end

    subgraph Model Layer [Model Layer]
        DataModels[Domain Data Models]:::model
        Repository[Repositories / Data Handlers]:::model
        AI[ONNX Runtime / WordPiece Tokenizer]:::model
    end

    subgraph Data Layer [Data & Local Storage]
        DB[(SQLite / sqflite)]:::data
        Cache[(Shared Preferences)]:::data
    end

    %% Flow of MVVM
    UI -->|User Intent / Actions| Riverpod
    Riverpod -->|Reactive State Updates| UI
    
    Riverpod -->|Fetch / Process Request| Repository
    Repository -->|Parsed Domain Data| Riverpod
    
    Repository -->|Query| DB
    Repository -->|Cache| Cache
    
    Riverpod -->|Search Query / Embeddings| AI
    AI -->|Semantic Match Results| DataModels

Loading

• View Layer: Contains the modular Flutter screens and reusable UI components. Responsible only for rendering state and capturing user input.
• ViewModel Layer (Riverpod): Acts as the bridge between the View and Model. It holds the business logic, manages the state of the UI, and interacts with repositories.
• Model Layer: Contains the core domain structures (Data Models) and the Repositories, which abstract the logic required to access data sources. Includes ONNX inference logic.
• Data Layer: Manages local persistence using SQLite for offline-first capabilities and Shared Preferences for caching.

2. Custom Input to Tags Pipeline

(Illustrates how user input is processed and mapped to semantic tags)

3. Recommendation Algorithm Overview

(Overview of rule-based filtering, injury exclusion, and semantic matching)

4. Voice Control Pipeline

(Flowchart detailing how voice commands are captured, processed, and executed)

---

🔄 User Flow

User opens the app
        ↓
User completes onboarding (details, goals, injuries)
        ↓
Algorithm & AI create a personalized micro-session
        ↓
User executes and completes the workout session
        ↓
User submits post-workout ratings and feedback
        ↓
System adapts and tailors the next session based on feedback
        ↓
User receives notification when the next customized session is ready

---

🍀 Getting Started

Prerequisites

• Flutter SDK
• Dart SDK
• Android Studio / Xcode (for emulation and building)

Installation

1. Clone the Repository

git clone https://github.com/AOSSIE-Org/MoveYourBody.git
cd MoveYourBody

2. Install Dependencies

flutter pub get

3. Generate Riverpod Code

dart run build_runner build --delete-conflicting-outputs

4. Generate Exercise Database

Run the provided Python script to set up the local exercise database.

python scripts/generate_exercise_db.py

5. Run the Application

Ensure you have a simulator running or a device connected.

flutter run

---

📱 App Screenshots

---

🙌 Contributing

⭐ Don't forget to star this repository if you find it useful! ⭐

Thank you for considering contributing to this project! Contributions are highly appreciated and welcomed, read the CONTRIBUTING.md for setting the project.

Note: Before opening a UI Pull Request, please ensure you read our brand.md file for styling guidelines.

---

📍 License

This project is licensed under the GNU General Public License v3.0. See the LICENSE file for details.

---

💪 Thanks To All Contributors

Thanks a lot for spending your time helping MoveYourBody grow. Keep rocking 🥂

© 2026 AOSSIE