AI Solutions2024
AI-Powered Recommendation Engine
A collaborative and content-based filtering recommendation engine delivering personalised product and content recommendations at scale.
Overview
A machine learning recommendation system built for an e-commerce client. The engine combines collaborative filtering (user-item matrix factorisation) with content-based filtering (product embeddings) to surface personalised product recommendations. Served via a FastAPI endpoint with Redis caching for sub-10ms latency.
Key Highlights
Hybrid collaborative + content-based filtering
Real-time personalisation at <10ms latency
A/B testing framework for recommendation variants
Explainability layer for recommendations
Automated retraining pipeline