E-Commerce REST API

A production-style REST API for an e-commerce platform built with Flask, SQLAlchemy, and JWT authentication. Features an ML-powered product recommendation endpoint using sentence transformers and cosine similarity.

Live API

Base URL: https://ecommerce-api-8zsy.onrender.com

Tech Stack

• Backend: Python, Flask, SQLAlchemy, Flask-Migrate
• Auth: JWT (Flask-JWT-Extended), bcrypt
• ML: Sentence Transformers (all-MiniLM-L6-v2), scikit-learn
• Database: SQLite (development), PostgreSQL (production)
• Deployment: Render

Features

• JWT authentication with role-based access (customer/admin)
• Product search and filtering by category and price
• Pagination on all list endpoints
• ML-powered recommendations using TF-IDF similarity
• Rate limiting on auth endpoints (5 requests/minute)
• Unit tested with pytest
• Deployed on Render with PostgreSQL

Setup

1. Clone the repo

git clone https://github.com/your-username/ecommerce-api.git cd ecommerce-api

2. Create virtual environment

python -m venv venv venv\Scripts\activate # Windows source venv/bin/activate # Mac/Linux

3. Install dependencies

pip install -r requirements.txt

4. Set environment variables

Copy .env.example to .env and fill in your values.

5. Run migrations

flask db upgrade

6. Start the server

python run.py

API Endpoints

Auth

Method	Endpoint	Description	Auth
POST	/auth/register	Register new user	No
POST	/auth/login	Login and get token	No

Products

Method	Endpoint	Description	Auth
GET	/products	List all products	User
GET	/products?category=electronics	Filter by category	User
GET	/products/:id	Get single product	User
POST	/products	Create product	Admin
GET	/products/:id/recommendations	ML recommendations	User

Orders

Method	Endpoint	Description	Auth
POST	/orders	Place an order	User
GET	/orders	Get order history	User
GET	/orders/:id	Get single order	User

Admin

Method	Endpoint	Description	Auth
GET	/admin/orders	View all orders	Admin
PATCH	/admin/orders/:id/status	Update order status	Admin
DELETE	/admin/products/:id	Delete product	Admin

ML Recommendations

The /products/:id/recommendations endpoint uses semantic similarity to find related products. Product names and descriptions are encoded using the all-MiniLM-L6-v2 sentence transformer model, and cosine similarity is used to rank results.

This means "noise cancelling headphones" and "audio device with sound isolation" would score high similarity even with no shared keywords.

Example Request

POST /auth/login Content-Type: application/json

{ "email": "user@example.com", "password": "password123" }

Response: { "access_token": "eyJhbGci...", "user": { "id": 1, "email": "user@example.com", "role": "customer" } }