Heat is one of the quietest dangers we face. It doesn't announce itself like a flood or a fire. There is no sudden event to warn you that something has gone wrong. Instead, heat builds gradually—one degree at a time, hour after hour—until a room that seemed perfectly comfortable can become an unsafe place to stay. And often, the people most vulnerable to that change are the least likely to notice it themselves.

An elderly parent living alone. A baby sleeping in a nursery. A pet left at home on a hot afternoon.

A conventional thermometer can tell you that a room is 35°C. But temperature alone doesn't tell the whole story. Humidity can make the same temperature feel significantly hotter, and prolonged exposure can turn an uncomfortable environment into a serious health risk.

We wanted to build something that could recognize that difference.

That idea became Heat Sentinel. It's a room-level heat monitoring and early-warning system designed to continuously watch the environment and notify someone when heat begins to cross into dangerous territory.

Instead of simply asking "How hot is the room?", Heat Sentinel looks at the bigger picture: how hot it actually feels, what risk level that represents, and how long the room has remained there.

Because a brief spike in temperature is not the same as spending several hours in sustained heat.

Heat Sentinel is built around that idea. It doesn't just measure the environment—it keeps track of how the conditions change over time, turning a simple temperature reading into a clearer picture of what is actually happening inside the room.

And most importantly, it doesn't require someone to constantly check a thermometer or keep a dashboard open. When conditions become dangerous, Heat Sentinel brings the warning to you. That is the gap we wanted to close: the gap between knowing the temperature and understanding the danger.

At the edge of the system is a small Waveshare ESP32-S3-Zero paired with a Waveshare BME280 sensor. The sensor measures temperature and humidity and sends that information over Wi-Fi to a Java 26 backend. From there, the raw measurements become something much more meaningful.

The backend calculates the Heat Index, using the established NOAA regression that accounts for both temperature and humidity to estimate how hot the environment actually feels. The result is then classified into different risk levels, allowing the system to understand not just whether a room is warm, but whether its conditions are becoming unsafe.

A brief spike in temperature is not the same as spending several hours in sustained heat. Heat Sentinel therefore keeps track of how long the room remains within each risk zone throughout the day, giving us a much clearer picture of the environment and its progression.

All of this information comes together in a live web dashboard, where the current temperature, humidity, Heat Index, risk level, and history can be monitored in real time. But we didn't want Heat Sentinel to depend on someone constantly watching a screen.

When conditions enter a dangerous range, the system can immediately send a Telegram notification to the person responsible for the space. The room doesn't have to wait for someone to check it—the system brings the warning to them.

The complete journey is surprisingly simple: BME280 → ESP32-S3 → Java 26 → Heat Index → Risk Analysis → Dashboard → Telegram Alert

We also wanted the software side of the project to be an engineering exercise in its own right.

The backend is built directly on Java 26, without Spring, Maven, or a collection of external frameworks. Modern Java features such as sealed interfaces, pattern matching, exhaustive switches, records, and virtual threads allow us to build the sensor API, web dashboard, real-time WebSocket layer, and alert system as a single self-contained application.

The result is a system that is small enough to understand, simple enough to deploy, and powerful enough to continuously monitor...