eSentry

System Architecture Overview

eSentry utilizes a modern full-stack architecture with clear separation between frontend and backend services. The application is built with a Next.js frontend communicating with a Node.js backend API, both interacting with a shared PostgreSQL database through Prisma ORM.

I designed the system with modularity in mind, allowing each component to focus on its specific responsibility while maintaining clean communication patterns between services. The frontend and backend communicate via RESTful API endpoints, enabling structured CRUD operations for user accounts, monitors, and all other application features.

Data Layer & Shared Schema

For database management, I created a shared Prisma Schema npm package that serves as a single source of truth for all database models and their relations. This approach makes database changes and migrations significantly easier—I simply update the package and both frontend and backend automatically receive the updated schema.

The Prisma ORM abstracts away complex SQL queries in favor of a data-centered JSON approach, making database interactions more intuitive and maintainable. Here's an example of how monitor creation works with Prisma:

1// Create the eSentry monitor from the user request data object
2const monitor = await prisma.monitor.create({
3  data: {
4    userId,
5    keywords: userReq.keywords,
6    excludedKeywords: userReq.excludedKeywords || [],
7    minPrice: userReq.minPrice,
8    maxPrice: userReqmaxPrice,
9    conditions: userReq.conditions || [],
10    sellers: userReq.sellers || [],
11    status: 'inactive',
12  },
13});

This code structure allows me to focus on the data relationships rather than writing complex SQL queries, significantly improving development speed and reducing potential errors.

Authentication & Security

The frontend implements Auth.js to provide a streamlined authentication experience with OAuth providers (Google and GitHub). This not only simplifies the login process for users but also enhances security by leveraging established authentication providers.

Security is implemented in layers throughout the application:

• Auth.js protects frontend API routes before requests even reach the backend
• Backend middleware authenticates requests again by verifying the user ID from frontend requests
• Rate limiting using Redis caching prevents API overload by tracking how many calls each user makes

This multi-layered approach ensures that only authenticated users can access sensitive functionality and prevents potential abuse of the system.

Monitoring & Queue System

The heart of eSentry's monitoring capability is built around an efficient job queue system. When a user activates a monitor, their request enters a BullMQ queue waiting for processing. These jobs run at fixed intervals to check for new eBay listings.

The monitoring process works by:

1. Pulling data from the eBay Search API using the user's specified monitor parameters
2. Caching search results for one interval, creating a comparison point
3. Comparing new search results against the cached data to identify new listings
4. Triggering email notifications through Amazon SES when new matches are found

This approach ensures reliable monitoring while maintaining reasonable load on both the backend server and eBay's API.

Deployment & Infrastructure

The backend is fully containerized using Docker, which provides consistent environments across development and production. The Docker setup includes not just the API service but also PostgreSQL and Redis services, creating a complete self-contained system.

For hosting, I deployed the backend container to a DigitalOcean VPS running Ubuntu. The application uses a domain structure that separates concerns:

• Frontend application:

  esentry.dev

• Backend API:

  api.esentry.dev

This separation with HTTPS implementation ensures secure communication between all system components. The containerized approach also makes scaling and updating straightforward, as I can deploy new versions with minimal downtime and consistent behavior.