multi-bank-coding-challenge

Full-stack real-time trading dashboard used for the Multi Bank coding challenge. This repository contains two main projects and supporting documentation.

Project layout

• client/trading-dashboard — Next.js frontend (UI, E2E + unit tests). See: client/trading-dashboard/README.md and client/trading-dashboard/README.md for developer instructions.
• server/market-trading-service — Node + TypeScript microservice that simulates market data and exposes a small REST API. See: server/market-trading-service/README.md.
• docs/ — architecture and project-scope documentation.
• docker-compose.yml & README.DOCKER.md — Docker setup for running both services together.

Quick start (recommended: Docker)

1. From the repository root, build and start both services with docker-compose:

docker-compose build
docker-compose up

This will start:

• Frontend: http://localhost:3000
• Backend API: http://localhost:3005 (health endpoint: /health)

Local development with Docker (hot reload)

Use the development compose file to run both services with live reload and all required env vars configured:

docker compose -f docker-compose.dev.yml up --build

This will start:

• Frontend (dev server): http://localhost:3000
• Backend API (dev): http://localhost:3005
• WebSocket (dev): ws://localhost:8080

Local development (without Docker)

• Frontend:

cd client/trading-dashboard
npm install
npm run dev

• Backend:

cd server/market-trading-service
npm install
npm run dev

Tests

• Frontend unit tests: run in client/trading-dashboard with npm run test (Jest)
• Frontend E2E: run Cypress after starting the frontend with npm run cypress:open for local npm run cypress:run for CI/CD pipelines
• Backend unit test run in server/market-trading-service with npm run test (Jest), npm run test:watch, npm run test:coverage also available if working on test and see the coverage

Assumptions & Trade-offs

Assumptions

• Data model: simplified ticker domain (price, change, volume, 24h high/low); no corporate actions, splits, multi-currency, or latency-sensitive guarantees.

• Simulation: price history and live ticks are synthetic for UX validation, not financial accuracy.

• Environment: single-node backend with in-memory storage is acceptable for this challenge; no cross-process persistence required.

• Client: no auth is required to view market data.

• Contracts: WebSocket message shapes are minimal and stable: {"type":"connected","payload":{"clientId"}} and {"type":"data","payload":{"ticker"}}. REST endpoints are reachable for initial bootstrap/fallback.

• CI/CD via github actions to run test on PR

• Used Code Rabbit to review

Trade-offs

• Real-time delivery vs simplicity: WebSocket for live updates plus a one-time REST bootstrap for fast first paint (slight duplication accepted for responsiveness).

• Subscription scope: initial subscribe-to-all for clarity; future optimization could subscribe only to visible/selected symbols.

• Update cadence: immediate per-tick broadcasts, no batching/debouncing; simple but more frames under heavy load. Batching window (e.g., 50–150 ms) can be added later.

• State storage: in-memory repository and subscription registry (per-process); not horizontally scalable without a shared store/pub-sub or sticky sessions.

• Consistency vs responsiveness: values rounded for display and updated on each data frame; suited for UI, not for reconciliation.

• Error handling: if REST fails, fall back to mock data; if WS drops, auto-reconnect and keep last-known data. Prioritizes resilience for demos.

• Client architecture: Next.js app with a WebSocket Provider context and React Query; SSR of live data is out of scope to keep the real-time logic client-side and testable.

• Testing: REST fallback and provider no-op defaults keep unit tests stable; deep WS integration tests are limited and can be added with a small WS mock.

• Selection model: selection tracked by symbol and derived from the latest ticker list to ensure live updates without stale references.

• Some of the Bonus features aren't covered due to time constrain.

License

This project is available under the repository LICENSE file.