When I entered college a few months earlier, I was the freshman tagging along with my upclass in hackathons.
This was different. For the first time, I was actually leading the team.
That meant doing more than writing code. I coordinated the project, assigned tasks, made technical decisions, set up structure so everyone could work in parallel, and eventually stood in front of the judges to present everything we’d built.
Build something people remember
The hackathon gave us complete freedom as long as the project related to transportation.
We could’ve built another navigation app. Another educational game. Another safe project. Instead, I pushed for a driver monitoring system that used computer vision to detect drowsiness and distraction through nothing more than a phone camera.
Some of my teammates weren’t convinced at first. It sounded difficult, maybe even impractical. My argument was simple: this is a hackathon. We don’t need an idea nobody has thought of. That’s almost impossible. We need an idea other teams think is impossible.
Three weeks of figuring things out
Choosing the ambitious project was the easy part. Actually building it wasn’t.
None of us had worked with computer vision or WebRTC before, let alone mobile apps. We spent the first weeks just getting the detection pipeline to work consistently.
Our biggest problem was performance. Our initial deployment to Render’s free tier didn’t work at all. We had to move to Azure App Service mid-development (because it had free credits for students) just to make live monitoring usable. Even then, a locally run backend was still faster, so we planned to use that for the demo instead.
The night before demo day, we panicked when the connection wouldn’t establish in the APK build. It turned out the fix was only a one-line config change after hours of debugging :)
The demo mattered (maybe the most)
When our turn came, we let the product do most of the talking. The judges watched the live detection. Then the map integration. Then the live tracking. Then recorded video analysis. Every time it looked like the demo was over, there was another feature waiting.
One judge commented that the project felt production ready (even though it isn’t lol). They also complimented how detailed our presentation was, especially the architecture diagrams we used (just used mermaid btw).
Looking back
This was my first hackathon outside campus, first time leading a hackathon team, and my first time winning champion.
Unlike the past hackathon, I don’t think we won because we used computer vision. We won because we built something people remembered.
Computer vision was the hook. The polished demo, the attention to detail, and the willingness to take on a project other teams avoided were what made the difference.
How it turned out
Champion out of 7 competing teams with our project Manobela.
Our project
We built Manobela, a driver monitoring system that detects unsafe driving behaviors in real time and alerts the driver immediately.
I also designed a pretty cool poster for the competition (bummer it didn’t win though).
