Code Collab

An AI-powered collaborative code editor with isolated code execution and real-time collaboration capabilities.

🎥 Demo Video

code-collab-demo-full.mp4

🚀 Features

• Real-time Collaborative Editing: Multiple users can edit code with cursor tracking simultaneously using Y.js and WebSockets
• Isolated Code Execution: Execute code in isolated environment safely using self-hosted Judge0 with AWS Auto Scaling groups and AWS SQS queue-based scaling
• AI-Powered Assistance: Get intelligent code suggestions and help via OpenAI API and interact with Jarvis AI via Chat Interface for real-time assistance
• WebSocket Server: Custom-built scalable WebSocket server for Chat feature with Redis pub/sub for horizontal scaling
• User Authentication: Secure user management with Clerk
• Modern UI: Clean, responsive interface built with Next.js and Tailwind CSS
• Scalable Architecture: Monorepo setup with Turborepo for efficient development with Docker and AWS Auto Scaling groups

Technology Stack

• Frontend: Next.js, React, TailwindCSS, shadcn/ui
• Real-time Collaboration: Y.js, CodeMirror, WebSockets
• Backend: Node.js, Express
• Database: PostgreSQL, Prisma ORM
• Execution Environment: Containerized Judge0 (self-hosted)
• Deployment: Docker, AWS Auto Scaling
• Scaling: Redis pub/sub for WebSocket scaling
• Authentication: Clerk

🏗️ Architecture

🛠️ Deployment Architecture

Current Setup

• Web App: A full-stack application built with Next.js, TailwindCSS, shadcn/ui, and Prisma ORM which has a frontend and backend. The frontend has features like creating rooms, real-time collaborative code editor, chat, and a dashboard.
• WebSocket Server: WebSocket server is highly scalable and can handle a large number of connections due to Redis pub/sub.
• Judge0: Self-hosted Judge0 running on AWS Auto Scaling groups with AWS SQS queue-based scaling. The system maintains around 10 messages in the queue, automatically scaling up instances when the queue grows larger and scaling down when the queue decreases.
• AWS CloudWatch: Monitors instance and queue metrics to facilitate the auto-scaling policy.
• Custom AMI: Created a specialized Amazon Machine Image with Judge0 and CloudWatch configurations for consistent deployment.

Scaling Infrastructure

• AWS SQS Queue: Code submissions are enqueued to SQS, which triggers scaling events based on queue length.
• Auto Scaling Policy: Configured to maintain approximately 10 messages in the queue at any time, adding instances when the queue grows and removing them when it shrinks.
• CloudWatch Integration: Each instance reports metrics to CloudWatch for monitoring and scaling decisions.

Monorepo Structure

code-collab/
├── apps/                 # Application packages
│   ├── web/              # Next.js frontend application
│   └── websocket/        # Custom WebSocket server
├── packages/             # Shared packages
│   ├── db/               # Prisma database schema and client
│   ├── ui/               # Shared UI components
│   ├── eslint-config/    # ESLint configuration
│   └── typescript-config/# TypeScript configuration
└── docker/               # Docker configuration files

AutoScaling Group Configuration

ASG.Creation.Steps.Demo.mp4

AutoScaling Demo

ASG.Demo.mp4