python-challenge.md

Task: Build a Real-Time Product Recommendation System

Duration: 3 hours

Objective: Assess the candidate's ability to design and implement a scalable real-time recommendation system for an e-commerce platform using Python, FastAPI, Celery, PostgreSQL, and machine learning techniques.

Task Breakdown

Part 1: Application Setup (30 minutes)

1. FastAPI Application:

• Set up a FastAPI application with a well-structured project layout.
• Configure necessary libraries such as FastAPI-SQLAlchemy for ORM and FastAPI-Cors for handling CORS issues.

2. Database Configuration:

• Connect to PostgreSQL and create tables for users, products, orders, and user_product_interactions (to track user interactions with products).
• Define relationships between tables.

Part 2: User Interaction Tracking (30 minutes)

1. Capture User Interactions:

• Implement an API endpoint that logs user interactions with products (e.g., views, clicks, purchases).
• Store these interactions in the user_product_interactions table.

2. Data Processing:

• Create a background task using Celery that processes the interaction data periodically to update user profiles for recommendations.

Part 3: Recommendation Algorithm Implementation (1 hour)

1. Collaborative Filtering Algorithm:

• Implement a simple collaborative filtering algorithm to generate product recommendations based on user interactions.
• Use techniques such as cosine similarity or matrix factorization (e.g., SVD) to compute recommendations.

2. Real-Time Recommendations Endpoint:

• Create an API endpoint that returns real-time product recommendations for a given user based on their interaction history.
• Ensure that the recommendations are updated dynamically as new interactions are logged.

Part 4: Advanced Features and Testing (1 hour)

1. Integration with Machine Learning Model:

• Integrate a pre-trained machine learning model (e.g., using scikit-learn or TensorFlow) that predicts user preferences based on historical data.
• Allow the model to be retrained periodically using new interaction data.

2. Unit Testing and API Documentation:

• Write unit tests for the recommendation logic and API endpoints using pytest.
• Document the API endpoints using Swagger or Postman.

3. Serialization Methods:

• Implement serialization of user interactions and recommendations using both pickle and JSON.
• Demonstrate converting data from pickle format to JSON for API responses.

Evaluation Criteria

• Code Quality: Clarity, organization, adherence to Python best practices.
• Functionality: Completeness of features implemented within the time limit.
• Algorithm Understanding: Ability to explain and implement collaborative filtering concepts effectively.
• Real-Time Processing: Efficiency in handling real-time data updates for recommendations.
• Testing Rigor: Completeness and effectiveness of tests written.
• Creativity: Innovative approaches taken in implementing features or algorithms.

References for Implementation

• For guidance on building FastAPI applications:
  • FastAPI Documentation
• To understand how to set up Celery:
  • Celery Documentation
• For collaborative filtering algorithms:
  • Introduction to Recommender Systems
• For machine learning with scikit-learn:
  • Scikit-learn Documentation