Project
Marquee
Intelligent movie discovery powered by custom ML algorithms.

Overview
The engine behind Marquee was built to solve the inaccuracy of generic search tools. It implements foundational machine learning algorithms, including SVD and Cosine Similarity, to identify deep patterns in user tastes and film characteristics across a dataset of thousands of movies and ratings.
The system uses a hybrid matching model that considers both genre categories and thousands of specific metadata keywords. It also includes a time-based decay function to prioritize results that align with your preferred eras of cinema, ensuring that every recommendation feels both relevant and insightful.
The platform features a visual dashboard with real-time fuzzy search and high-resolution poster previews. By integrating live streaming data and IMDb links, it provides a comprehensive experience that helps film enthusiasts move directly from discovery to viewing in a single interface.
Stack
Python, Streamlit, Scikit-Learn, Pandas, TMDB API, NumPy, Requests
Status
Complete