Reel Rizz

Dating apps ask you to describe yourself, but how you actually spend your time, the reels you rewatch at 2am, the niche corners of the internet you've fallen into, that's the real you. Reel Rizz skips the bio and matches people on what they actually watch. Upload a screenshot of your Instagram Explore page's interest bubbles, we run OCR to extract your taste tags, embed each one semantically, and average them into a single "vibe vector." Then we rank every other user by cosine similarity -- the closer the angle between your vectors, the higher your compatibility. From there, you browse a TikTok-style feed of your top matches' latest Instagram reels. Like 3 reels from someone and you've liked them. Mutual likes unlock each other's Instagram handle, no in-app messaging, no ghosting limbo, just a real connection on a real platform. Built with TypeScript, React, and Node.js on the frontend and backend, a touch of Python for the embedding pipeline, and MySQL via Prisma for persistence.

Built With

• mysql
• node.js
• prisma
• python
• react
• typescript