Health-focused beverage recommendation engine tailoring suggestions to user profiles, dietary restrictions, and taste preferences. Built rule-based system allowing nutritionists to author transparent recommendations without requiring ML expertise or large datasets. Redis caching keeps suggestion latency low.
🎯 Overview
Chooze recommends beverages aligned with health goals through explainable rule-based matching. Users understand why each beverage is recommended, building trust that black-box ML models cannot achieve. Nutritionists tune rules via configuration without engineering involvement.
🛠️ Tech Stack
- Node.js: Shared JavaScript runtime between frontend and backend, enabling schema reuse and reducing integration surface.
- TypeScript: Strict typing on recommendation rules and user profiles preventing silent attribute-mismatch bugs.
- PostgreSQL: Relational storage of user health profiles and preference history supporting multi-attribute filtering.
- Redis: Pre-computed recommendation sets by profile hash eliminating repeated rule evaluation.
📈 Key Features
- Explainable rule-based recommendations transparent to health-conscious users
- Nutritionist-tunable rules without engineering deployment cycles
- Multi-attribute health profile matching against configurable guidelines
- Cached suggestion sets for common profile combinations
- Feedback loops for rule refinement based on user interactions