Real-time environmental monitoring for safety

The challenge

The client had to keep outdoor structures safe depending on the weather, but forecasts have a resolution of kilometres and can’t see a single spot. They needed reliable local measurement, without promising what a software platform cannot guarantee.

How we worked

• Requirements analysis with the client and a written feasibility study: what is included, what is excluded by design and what must be verified before a binding offer.
• Research into available measurement hardware and a comparison of kit configurations, with an initial module dedicated to validating them in the field.
• A modular development plan, each module with scope, acceptance criteria and a fixed price, and the option to pause at any time without losing work.
• Responsibility boundaries set from day one: the platform measures and alerts, it does not judge structural fitness.

Architecture and choices

• Three independent data paths: real-time display, threshold evaluation and storage. A slow archive can never delay an alarm.
• Tenant isolation at device level: every kit has its own credentials and cannot publish another client’s data.
• An alarm engine with per-position thresholds, warning and alarm levels, hysteresis and minimum exceedance time, to avoid false alarms.
• Raw data written once, timestamped at measurement rather than reception, so history can be used as evidence.

External services, and why these

Outcome

Study delivered with per-module estimates, recurring costs listed separately and risks made explicit. The client got a document to decide on, not a blind quote.

Stack

• AWS IoT Core
• MQTT
• AWS Lambda
• Time-series storage
• React Native
• Push notifications
• Weather APIs