LevelZ | Synchronized Codelab

Live Multiplayer Quiz Arena

Project Overview

LevelZ (levelz.co.in) is a multiplayer quiz gaming platform where users join subject-based real-time quiz games and compete on live leaderboards. Top 3 players win prizes. You built scalable FastAPI backend with OpenAI integration for dynamic question generation, React admin panel for game management, AI bots for simulated gameplay, and BullMQ automation for scheduled batch processing with uniqueness checks and caching.

Industry : Gaming & EdTech

Location : India

Year : 2025

Technologies Used

React.js

Node.js

OpenAI

Express.js

Stripe

FastAPI

AI Bot

MongoDB

React Native

Challenges

LevelZ's real-time multiplayer setup faced scalability issues during peak hours with hundreds of simultaneous users. Low-latency leaderboards, score updates, and WebSocket sync needed heavy optimization, while AI question generation required deduplication for thousands of unique items and robust AI bots handled session dropouts.

Real-time Scalability: Tuned FastAPI WebSockets + Redis for 1000+ users, latency <50ms.
Session Reliability: Atomic transactions fixed score disputes in dynamic games.
Automation: BullMQ retries achieved 95% reliable scheduling.

Solutions

1. SCALABILITY

High concurrent traffic caused latency spikes and unstable real-time performance. Deployed Redis pub/sub with FastAPI WebSockets to support 1000+ concurrent users, reducing latency from 500ms to under 50ms during peak loads.

2. AI DEDUPLICATION

Duplicate content in large question datasets impacted quality and user experience. Implemented OpenAI embeddings with MongoDB vector search to achieve 99.8% uniqueness across 50K+ questions, while improving processing speed by 3x using batch operations.

3. AUTOMATION

Unreliable job processing and failures in task scheduling affected system stability. Integrated BullMQ with dead-letter queues, retry mechanisms, and health checks, achieving 95% reliability in scheduling with automated fallback handling.

Results

LevelZ delivered production-ready scalability, handling 1000+ concurrent users at <50ms latency (95% improvement from 500ms), generated 50K+ unique questions with 99.8% deduplication at 3x batch speed, achieved zero score disputes across 500+ games via atomic transactions, and cut manual overhead by 90% through 95% reliable BullMQ automation with AI bots.