Not Another Rewatch
Semantic movie search backed by offline embeddings.
What
A movie-discovery app built around semantic search and taste-based recommendations. It combines similar-movie lookup, ratings, watchlists, achievements, and optional chat without making the core experience depend on a hosted LLM.
Approach
An offline Python pipeline builds sentence-transformer embeddings and writes the catalog to DynamoDB. A Java Spring Boot API serves search and recommendation requests to a React client. LocalStack and Docker Compose keep the full data path reproducible during development.
Stack
Java 21, Spring Boot, React, TypeScript, DynamoDB, Python, sentence-transformers, LocalStack, and Docker Compose.