IoT Machine Fleet Dashboard
by: Tomás Ospina.
Overview
A production monitoring platform for game machines installed across different locations. Each machine carries an ESP32 with a distance sensor that reports how full its coin chest is; the readings travel over MQTT to a NestJS backend and land in a React dashboard where the operator sees every site at once, with history and trends.
Problem & Solution
Checking a machine meant physically driving to it. The fix was to make the chest report its own state: an ultrasonic (HC-SR04) or IR sensor measures the fill level, the ESP32 publishes it over MQTT/TLS, and the backend subscribes with a wildcard topic so adding a machine requires no code change — it simply starts publishing and appears in the dashboard.
Technical Specifications
| Layer | Detail |
| ———- | ————————————————————- |
| Firmware | ESP32 DevKit v1 + HC-SR04 / FC-51 / TCRT5000, PlatformIO |
| Transport | MQTT over TLS — HiveMQ Cloud in production, Mosquitto locally |
| Backend | NestJS, REST + WebSocket, Swagger at /api/docs |
| Storage | PostgreSQL, historical series per machine |
| Frontend | React dashboard with role-based users |
| Infra | Docker Compose (PostgreSQL + Mosquitto) |
My Role & Key Contributions
- Designed the full stack, from the sensor wiring to the dashboard.
- Wrote the ESP32 firmware and made the publish interval configurable so the same device can serve short-term debugging and long-term trend collection.
- Built the NestJS ingestion service, the PostgreSQL schema for historical series, and the REST/WebSocket API.
- Implemented the React dashboard with historical charts and per-role access.
- Packaged infrastructure with Docker Compose so the whole stack comes up in one command.
What I Learned
The wildcard MQTT subscription was the design decision that paid off most: it turned “deploy a new machine” from a backend task into a purely physical one.
🔒 Private repository — happy to walk through the code on request.