This is a submission for the GitHub Finish-Up-A-Thon Challenge
What I Built
I built CineMatch, an AI-powered movie recommendation platform designed with a cinematic Netflix-inspired interface.
The project started as a small recommendation engine, but during this challenge I focused on transforming it into a much more polished and immersive experience. I redesigned the UI, improved the recommendation logic, added responsive layouts, integrated TMDB live metadata, and reorganized the project architecture into modular components.
The platform uses TF-IDF vectorization and cosine similarity to recommend movies based on storyline similarity instead of only genres.
Some major features include:
- AI-powered semantic movie recommendations
- Typo-correcting search system
- Live TMDB integration
- Dynamic spotlight movie banners
- Responsive mobile-first layouts
- Personalized watchlist system
- Glassmorphism cinematic UI
- Mood-based recommendation steering
The app currently works with a local database of more than 45,000 movies.
Demo
Live Demo
https://cinematch-movie-recommender-bqzsppgvepmahksda8qxg9.streamlit.app/
GitHub Repository
https://github.com/codewithvishuuu/cinematch-movie-recommender
Screenshots
| Home Landing | AI Matcher | Responsive Layout |
|---|---|---|
 |
 |
 |
The Comeback Story
Originally this project was a very simple recommendation prototype with basic UI and minimal functionality.
During the challenge I completely reworked large parts of the project. I reorganized the structure into modular folders, improved the frontend design, added reusable UI components, optimized mobile responsiveness, and connected live TMDB APIs for trailers and metadata.
One of the biggest improvements was making the interface work properly on smaller screens. Earlier many sections broke on mobile devices, especially buttons, grids, and text layouts. I redesigned the layouts using responsive spacing, adaptive sizing, and flexible containers.
I also improved the recommendation engine by adding better search handling, typo correction, and mood-based filtering.
The final result feels much more like a complete product instead of just a small ML experiment.
My Experience with GitHub Copilot
GitHub Copilot helped me speed up repetitive development tasks and UI restructuring.
I mainly used it while:
- reorganizing components
- improving responsive CSS
- generating repetitive layout sections
- refactoring utility logic
It helped reduce development time, especially during UI polishing and architecture cleanup, while I still manually customized the recommendation logic and overall design decisions.
Overall this challenge helped me improve both my frontend design skills and project organization workflow.
Top comments (0)