Back in February, there was this huge storm — like heavy rain, crazy winds, nonstop thunder. It was called Yagi, and it caused a lot of trees to fall over, damaging roads and even leading to some deaths. What really shocked me was that even after the storm ended, trees kept collapsing because their roots were already weak or the soil was too loose. That’s when my friends and I started thinking: “What if there’s a way to monitor trees and warn people before this happens?” And that’s how this idea began. 

At first, our goal was to display everything on a website. But here’s the problem: the site only worked on LAN (local network), which means only devices connected to the same Wi-Fi could see it. Not practical at all. So, we tried to explore other options — Java, JavaScript, CSS, and HTML — but honestly, they were too hard for us. None of us had ever touched those before, and learning them from scratch would’ve taken way too long.

That’s when I suggested using MIT App Inventor. I had learned it back in secondary school, and it’s super beginner-friendly — kind of like Scratch, where you just drag and drop blocks to build stuff. Since I already knew how it worked, I was assigned to design the app.  I made it connect to ThingSpeak, an online platform that stores our sensor data. This way, the app could pull live data from the cloud and display it on any phone, anywhere — no LAN problems! 

This was also how we divided tasks as a team: everyone took on what they were best at or most comfortable with. We each had different strengths, and we made it work together.

We met each other in an online coding class and decided to join this competition as a team. We had already learned about AI, so we divided the work based on each person's strengths—AI, sensors, mobile app, and hardware. The hardest parts were definitely the physical circuits and the coding. There was a lot of code, and it had to work smoothly with the real-world sensors and wiring, so syncing everything was quite a challenge. But thanks to teamwork (and a lot of testing), we made it through!

At the competition, I faced my first challenge: soldering a single sensor. What should have taken a few minutes stretched into half an hour of shaky hands, trial, and error. But that struggle was where I learned the most. By the next parts, I was faster, steadier, and more confident. That 30-minute battle turned into the foundation for the skills I carried through the whole build. 

At first, some teammates wanted to use Java since they were familiar with it, but I preferred Python. In the end, we decided to use C++ for the sensors and Python for the AI parts—which meant I had to learn some C++ too!

We met monthly on Google Meet from January to late April. Usually, we worked on our own for a week, then asked for feedback or help from our mentor if needed. Thankfully, with our mentor’s support, we didn’t face any big issues—we always had a way to solve things.