Customer Support

The Customer Support Center wants to add, edit and remove Support Agents. Customers want to be able to report a case for returning a product. Your software should assign an available Support Agent to handle a case automatically Support Agents should be able to list and resolve all the current active cases and Customers can add a new case. An agent can have only one case at a time. When the case is resolved, it should be marked so and the Support Agent should be free to take a new case

Install

Pre-Requisites

Before starting with system installation, the following are required

• Docker
• Docker Compose

Steps

1. Clone the repository.
2. Copy the .env.example file to .env: cp .env.example .env
3. pdate the values in the .env file (if needed).
4. Run make install (For initial setup)
5. Run the application make run (To run the containers)
6. Run make shell service=node cmd="npm run test"
7. Run docker exec -it support-node bash
8. Run npm run start:dev"

Note: Seed data is not available, so testing must be done using Postman for creating Customers and Agents, and the front-end for creating, listing, and resolving tickets.

Test steps:

• Create customer [Postman]
• Create Agent [Postman]
• Login as Customer [Frontend]
• Create Ticket [Frontend]
• Create as agent [Frontend]
• Resolve Ticket [Frontend]

Technical Details

Technologies:

1. Nest.js.
2. Mysql database.
3. Typeorm as ORM.
4. Jest

Architecture:

I used Monolithic Modular

My assumption

• I assumend this is as POC or MVP
• Scalability: I assumed the scale is not so high and I can handle from the infrastructure in this stage
• I don't give the security attention in this POC

Design:

1. Repository for accessing the database layer.
2. Service Layer has business logic.
3. DTO pattern
4. I tried as I can follow framework design recommendations.

Tradeoffs:

• Auto-assign ticket:
  • The availability of each agent is tracked using a flag and start time.
  • The ticket is assigned to the earliest available agent by pushing it onto a queue, and a job listens on this queue to pick the next available agent.
  • If no agent is available at the time, a warning will be logged and the job will retry three times before deleting the ticket from the queue.
  • A cron job will run every 5 minutes (based on the severity of the issue) to attempt to assign the ticket to an available agent.
• Relational database is used because the business requirements support this approach, with 4-5 entities having relationships with each other.

TODO:

• Authorization
• Validation "Specially the unique constrain and make sure order is belong to the customer"
• Using config service
• Implement filters in getTickets
• Throw custom errors to make tracing and monitoring more easier
• Cron job to assign tickets
• Ticket workflow "I have to validate before resolve the ticket that it is already in progress for example"
• Complete CRUD operation for agents controller
• Customers module
• Order module or at least order table in DB
• Database Seed
• CI/CD