Aarini Architecture

Aarini is a cross-platform mobile application built to provide health insights, cycle tracking, and ML-powered predictions. The system is split between a React Native (Expo) frontend and a Python-based backend handling the predictive logic, all bound together by Firebase for real-time data synchronization.

?? High-Level Overview

The architecture is divided into three primary tiers:

Client / Frontend (frontend/): An Expo React Native application providing the user interface, state management, and device integration.
Backend API (backend/): A Python-based service responsible for machine learning algorithms, cycle predictions, and data analysis.
Database & Auth (Firebase): Firestore provides real-time NoSQL data storage (governed by firestore.rules), while Firebase Authentication secures user sessions.

?? System Architecture Diagram

graph TD
    User[End User (Mobile)]
    
    subgraph Frontend Application
        Expo[React Native / Expo App]
        Context[React Context / State]
        Screens[UI Screens & Components]
        Services[API Services / Hooks]
    end
    
    subgraph Backend Services
        Python[Python API]
        ML[Cycle Prediction Engine]
    end
    
    subgraph Firebase Infrastructure
        Auth[Firebase Authentication]
        Firestore[Cloud Firestore DB]
    end
    
    User -->|Interacts| Screens
    Screens -->|Dispatches| Context
    Context -->|Uses| Services
    
    Services -->|Authenticates| Auth
    Services -->|Reads/Writes| Firestore
    Services -->|Fetches Predictions| Python
    
    Python -->|Queries History| Firestore
    Python -->|Runs Models| ML

?? Directory Structure

Frontend (`frontend/`)

assets/: Static images and fonts.
components/: Reusable UI components.
screens/: Full-page views for navigation.
context/: Global state management (Auth, Theme).
services/: API wrappers and Firebase interactions.
utils/: Helper functions and parsers.
navigation/: React Navigation configurations.

Backend (`backend/`)

app.py: Main application entry point for the REST API.
cycle_prediction.py: Core algorithm for predicting cycles.
tests/: Unit tests for backend logic.

?? Data Flow

User Input: A user logs a new symptom on the mobile app.
Frontend Service: The frontend/services layer packages this data and sends it securely to Firebase Firestore.
Backend Trigger: When the user requests a new prediction, the Python backend queries the recent data from Firestore.
Processing: cycle_prediction.py processes the historical data.
Response: The backend returns the predicted dates to the frontend, which updates the React Context and re-renders the UI.

?? Deployment Strategy

Frontend: Distributed via Expo Application Services (EAS).
Backend: Containerized/Deployed via standard PaaS providers (e.g. Render, Vercel, Heroku) as defined by render.yaml.
Database: Serverless deployment on Google Cloud via Firebase.

FilesExpand file tree

ARCHITECTURE.md

Latest commit

History