Tailor

Transparent Social Video Network

About This Project

Tailor is a technically complex social video platform built on a microservices architecture. The system uses Qdrant vector database for semantic content search and recommendation, Kafka for event streaming between services, and Redis for caching and real-time data. I designed the entire backend infrastructure to handle video processing, content delivery, and personalized feed generation at scale. The recommendation engine uses embeddings and vector similarity search to provide transparent, user-controllable content curation.

Key Highlights

• Architected microservices system with Kafka event streaming
• Implemented vector similarity search using Qdrant for recommendations
• Built video processing pipeline with FFmpeg for multiple formats
• Designed transparent recommendation algorithm with user controls
• Optimized Redis caching strategy for sub-10ms API responses
• Created efficient video streaming with adaptive bitrate

Challenges & Solutions

• Balancing algorithm transparency with user experience
• Optimizing video streaming performance
• Building scalable recommendation system