ML-Driven User Personalization Engine

I’m building a system that dynamically adapts our site to each visitor, showing the most relevant articles based on how they actually behave. To make this happen I need an end-to-end personalization pipeline that:

• captures clicks & page views, time on page, and onsite search terms,
• preprocesses that stream for modelling,
• trains and evaluates collaborative filtering, clustering, and content-based models in Python (scikit-learn or TensorFlow are fine),
• serves real-time recommendations through an API or micro-service, and
• refreshes the models on a predictable schedule as fresh data comes in.

I already have analytics hooks and a basic data store; what’s missing is the learning layer and the integration logic that will replace our current “one-size-fits-all” article feed. Clear, well-commented code, a short README, and a simple test harness that proves the engine personalises articles better than a popularity baseline will be my acceptance criteria.