Client storyIoT & cloud

Building the Server Side: An IoT Backend for Sensor Data and Live Camera Streams

A client had working devices but nowhere reliable to send their data. I set up the full backend: device connectivity, data storage, dashboards and live camera streaming.

~1,000devices supported
< 1 s~live video latency
1team owning device to dashboard
Client
IoT product company (under NDA)
Devices
Sensor nodes and camera units
Built
Ingestion, storage, dashboards, video streaming
Bonus
Firmware and server designed together

The problem

The client's devices worked on the bench, but the data went to a quick prototype script on a single server. Messages were lost when the server restarted, video only worked on the office network, and nobody could see what the fleet was doing.

How I solved it

  1. Device connectivity

    A secure MQTT broker with per-device credentials, so every device authenticates and can be revoked individually.

  2. Data pipeline

    Sensor data flows into a time-series database, with retention rules so storage costs stay predictable.

  3. Live video

    Camera units stream to a media server that delivers low-latency video to browsers and recordings to storage.

  4. Dashboards and alerts

    Live dashboards for the fleet, and alerts when a device goes offline or a reading goes out of range.

  5. Designed with the firmware

    Because I also worked on the firmware, the message formats, reconnection behaviour and buffering on both sides were designed together.

What the client got

Results

  • A backend that scales to around a thousand devices.
  • No lost messages when servers or networks restart.
  • Live video from anywhere, not just the office.
  • Dashboards and alerts the whole team can use.
Related guideDesigning an IoT Backend: MQTT, a Time-Series Database and Live Video Streaming
Nikhil Robinson

Embedded systems consultant at TechProgeny, Bengaluru. 7+ years and 29+ shipped products across firmware, embedded Linux, edge AI, apps and IoT servers.