Next.js • Node.js • TypeScript • Go • PostgreSQL • Redis • Docker • Kubernetes • RabbitMQ • Linux

github

Rohit Tomar  • Peerlist

Production-grade ecommerce backend built as a distributed microservices system. Services include API Gateway, Auth, Product, Order, Payment, Notification, Invoice, and File Storage. Key engineering decisions: 1. Designed an asynchronous event-driven architecture with RabbitMQ. Payment intent creation, invoice generation, and email notifications run independently of the request lifecycle, allowing core business operations to complete without waiting on external services. Stripe payment intents are typically generated before the user reaches checkout. 2. Implemented the Sidecar Pattern for payment processing. A dedicated worker consumes order events and handles all Stripe interactions independently of the Order Service. 3. Rebuilt the Invoice Service in Go after the initial Node.js implementation. Processing a backlog of 10,000 invoices through RabbitMQ improved from 1 minute 47 seconds to 2.43 seconds using goroutines. 4. Validated RabbitMQ durability by publishing 10,000 events while consumers were offline. All messages were successfully processed after service recovery with zero message loss. 5. Applied the Database-per-Service pattern, ensuring each service owned its own data and eliminating cross-service database dependencies. 6. Built and deployed the infrastructure from scratch using Nginx, Docker, Kubernetes manifests, and SSH-based deployment on a self-configured VM. 7. Integrated Stripe payments, S3-compatible object storage, Redis caching, and RabbitMQ eventing throughout the platform. Stack Node.js · Go · Next.js · PostgreSQL · RabbitMQ · Redis · Docker · Kubernetes · Nginx · Stripe · S3 GITHUB - https://github.com/tomarrohitt/e-commerce Live - https://rohit-ecommerce-microservice.dedyn.io

Ecommerce Microservice System

Full-Stack Multiplayer Chess Platform Built and deployed a production-grade multiplayer chess platform with real-time gameplay, a custom-tuned chess engine, in-game chat, inbox messaging, game archives, and move-by-move replay. Technical Highlights 1. Implemented resilient WebSocket connections with exponential backoff reconnection (5 attempts) and server-side reconnect event handling, allowing players to rejoin active games without losing state. 2. Stored active game state in Redis and persisted completed games to PostgreSQL. Managed frontend state with Zustand. 3. Enforced double-layer move validation using chess.js on both client and server. Invalid moves are rejected and corrected server-side. 4. Optimized chat performance through batched database writes. Messages are delivered instantly via WebSockets and persisted every 2 seconds, reducing database write pressure while maintaining real-time UX. 5. Used Web Workers for move calculations and timer updates, preventing main-thread blocking and ensuring a responsive UI during complex positions. 6. Leveraged Node.js Worker Threads for CPU-intensive tasks, keeping the event loop responsive under compute-heavy workloads. 7. Applied rate limiting across both HTTP APIs and WebSocket connections. 8. Integrated BullMQ for background job processing and Redis-based caching throughout the application. 9. Implemented secure cookie-based authentication using better-auth. Stack Next.js, Node.js, TypeScript, Redis, PostgreSQL, BullMQ, Zustand, chess.js, better-auth Deployment Frontend: Vercel Backend: Hugging Face Link - https://rohit-chess-app.vercel.app