Subtitle translation API using GPT. Dynamic context windows preserve conversational flow. Features auto-fallback to DeepL, concurrent processing, and FastAPI.

A nice FastAPI Reranker as a service

This project demonstrates how to build a semantic search system using Qdrant vector database. Startup descriptions are converted into embeddings and stored as vectors, while structured metadata is stored as payload.A curated demo dataset is used to clearly showcase vector search behavior and payload-based filtering.