ML RECOMMENDATION
CineMatch AI — Movie Recommendation Engine
ML-powered movie recommendation system. Search any movie title and instantly receive a ranked list of similar films computed via cosine similarity over TF-IDF content vectors. Live API on Render, React frontend on Netlify.
Public demo
10 — top recommendations per query
The problem
Streaming platforms bury similar films behind engagement-optimised carousels. A clean recommendation engine that tells you 'if you liked X, here are 10 films statistically closest to it' is more honest and useful.
The approach
Constructed a TF-IDF matrix over movie metadata (title, genre, plot keywords, cast). Precomputed a full cosine similarity matrix at build time so recommendations are O(1) lookups at runtime. FastAPI exposes a /recommend/?movie= endpoint. React + Tailwind frontend with real-time search, deployed on Netlify with the backend on Render.
The results
Live API serving recommendations for thousands of movies. Sub-50ms response time after model load. React frontend with glassmorphism UI and instant search. Publicly accessible at the Netlify URL.
Stack: React 18, Vite, Tailwind CSS, FastAPI, Scikit-learn, Python