CineMatch
Find a film you both want to watch
Two people join a room and swipe through films independently. When both like the same movie, a real-time notification makes the decision easier.

The problem
Two people can have an idea of what they want to watch but still cannot agree on a film. Each person's preferences leave them with different choices, and neither knows which films the other would also enjoy.
The solution
CineMatch lets two people join a shared room, select genres and reference films, and like films independently. Matching is deterministic: when both like the same film from TMDB, Socket.io notifies them in real time and the match stays in their history. The current flow does not use an LLM; AI recommendations remain planned, not implemented.
My responsibility
The documented project scope covers the React and TypeScript interface, a Node.js and Express backend, and real-time matching with Socket.io. This account does not break down individual and team responsibilities or establish sole ownership.
Constraints
- Room access uses short-lived sessions and a six-character code, without a long signup flow. Temporary sessions use JWTs.
- Each person swipes independently, but a shared like needs to notify both participants and remain available in match history.
Technical decisions
Connect without a long signup
Start a short-lived session with a username, create a room, and share its six-character code. Each person can select genres and reference films before browsing.
Match in real time
Users swipe independently through films from TMDB. Socket.io notifies both people when they like the same film, and match history keeps their shared choices available.
Build across the stack
A React and TypeScript frontend connects to a Node.js and Express backend. PostgreSQL and Prisma store the app data, with JWT-based temporary sessions.
Planned: AI recommendations
A future version could combine both users’ film preferences, concepts, and reference movies. An LLM, supported by retrieval from a film catalog, would suggest films that fit both people. This feature is not implemented yet.
Outcomes and evidence
- The implemented flow includes room creation, genre and reference-film selection, independent swiping, real-time match notifications, and match history.
- AI recommendations are planned, not implemented. The current matching flow uses shared likes on films from TMDB.
The existing project notes describe these capabilities, but do not report usage or match-rate metrics. Source code is linked below; no live demo is linked.