CineMatch AI
Movie Recommendation System
The Problem
Traditional movie recommendation systems rely on basic keyword matching or collaborative filtering, often missing the nuanced preferences users have. Users struggle to find movies based on "vibes," themes, or emotional tones they're looking for.
My Role
Full-Stack Developer & AI Engineer
The Solution
Built a hybrid recommendation engine that combines semantic AI with traditional filtering. The system uses sentence transformers to understand movie descriptions semantically, stores embeddings in Pinecone Vector DB, and provides a VOD-style interface for seamless discovery.
Technology Stack
Next.js
Frontend framework for server-side rendering and optimal performance
TypeScript
Type-safe development for maintainable code
Tailwind CSS
Utility-first styling for responsive design
Framer Motion
Smooth animations and transitions
FastAPI (Python)
High-performance backend API for ML inference
Pinecone
Vector database for semantic search capabilities
Sentence Transformers
NLP embeddings for understanding movie descriptions
TMDB API
Movie data, metadata, and poster images
Key Features
- •Semantic search that understands natural language queries like "dark sci-fi with philosophical themes"
- •Hybrid recommendation combining AI-powered semantic matching with traditional filters
- •VOD-style interface with smooth animations and intuitive navigation
- •Real-time movie suggestions based on user preferences and viewing history
- •Advanced filtering by genre, year, rating, and more
Impact & Results
Created a recommendation system that goes beyond keyword matching, allowing users to discover movies based on themes, moods, and vibes. The semantic search significantly improved user satisfaction and discovery rates.
Screenshots
CineMatch AI Homepage
Semantic Search Results
Movie Details & Recommendations
Lessons Learned
- •Learned to optimize vector database queries for sub-second response times
- •Gained experience in balancing AI complexity with user-friendly interfaces
- •Understood the importance of hybrid approaches combining multiple recommendation strategies
- •Mastered the integration of Python-based ML services with Next.js applications