Senior Backend Engineer Dreemz
Designed and scaled a real-time chat system for more than 25,000 concurrent users and 500,000 registered users. Built a horizontally scalable Socket.io cluster using a Redis adapter, enabling consistent session routing, low latency, and high availability across replicas.
Improved backend throughput by 60%, reducing a Lambda workflow from 16–21 seconds to 1–2 seconds through service-boundary optimization, efficient data-access patterns, and database tuning. Redesigned PostgreSQL indexing and query strategies, reducing p95 API response times from over 800 ms to under 400 ms.
Developed an Amazon S3 pre-signed URL upload pipeline that eliminated server-side file proxying and reduced server load by approximately 80–90%. Stabilized an AI-powered conversational platform using NestJS, AWS Bedrock, Redis, and Socket.io by resolving LLM orchestration, state-transition, and real-time streaming issues.