NearBuy

NearBuy is a shopping lists mobile app that sends location-based alerts when you're near stores that likely have items from your shopping list, with smart items recommendations based on community trends.

The system consists of three main components:

1. Model Service: A machine learning API that recommends product and predicts item availability based on store data.
2. Backend Service: A FastAPI application handling user lists, notifications, and integrating with Supabase.
3. Frontend App: A React Native (Expo) client for users to manage lists and receive location alerts.

📎 Additional Resources

• 📽 View the NearBuy Presentation
• 📄 Full Project Documentation (PDF)

---

Architecture Overview

+------------+        +-------------+        +-------------+
|  Frontend  | <--->  |   Backend   | <--->  |  Model API  |
|  (Expo)    |        |  (FastAPI)  |        | (uvicorn)   |
+------------+        +-------------+        +-------------+
        |                    |                      |
        v                    v                      v
    Expo Go / Dev        Supabase DB           Model Weights

• The Frontend communicates with the Backend over HTTP.
• The Backend stores data in Supabase and calls the Model API for recommendations.
• The Model API loads a trained model and serves predictions.

---

Prerequisites

• Node.js (v14+)
• Expo CLI: npm install -g expo-cli
• Python (v3.10+)
• pip or pipenv (for Python dependencies)

---

Getting Started

Navigate to the project’s root

Install dependencies

pip install -r requirements.txt

1. Local URL for Emulator / Device

Run this to get your local IP:

for /f "tokens=2 delims=:" %a in ('ipconfig ^| findstr "IPv4"') do @echo %a

in NearbuyFE/utils.tsx, change the currentPath variable at the beginning of the file like so:

// choose current path to match the emulator (and change to your IP)
export const currentPath =
  //'http://172.30.124.49:8001/' // for android emulator (remote IP)
  // 'http://localhost:8001/' // for web
  'http://<your_local_ip>:8001/' // for phone via USB / expo go app
  ;

2. Configure Environment Variables

Paste the .env file in the root directory.
Also, paste the firebase-service-account.json file in the NearBuyFE directory.

Model Service

This service loads the ML model and exposes endpoints for predicting item availability.

# Start the API
uvicorn ml_component.api.model_fastapi_server:app --reload --host 0.0.0.0 --port 8000

Backend Service

The FastAPI backend handles list management, user profiles, and notifications.

# Start the app
cd NearbuyBE/app
uvicorn main:app --reload --host 0.0.0.0 --port 8001

Frontend App

The React Native app built with Expo.

Notifications are currently available only in Expo Dev on Android due to cost restrictions.

# Navigate to frontend folder
cd NearBuyFE

# Install dependencies
npm install

# Start Expo
npm run android

• Scan the QR code for phisical mobile or press a to open in an emulator.
• Ensure the Backend (localhost:8001) and Model API (localhost:8000) are running.

---

Running the Full Stack

1. Model Service (port 8000)
2. Backend Service (port 8001)
3. Expo Frontend (Expo CLI)

In separate terminals or via a process manager (e.g., tmux, Docker, or pm2), start each component in order. Then open the Expo app and log in or sign up to begin.