- Problem
- Mosquito-borne diseases such as dengue remain a persistent public health challenge, and identifying breeding areas early is difficult with manual monitoring alone.
- Solution
- A mobile and cloud system that uses computer vision to help identify the mosquito and map reports geospatially so that data can support prevention efforts.
- My role
- Co-founder, AI and software engineer
- Result
- Global Winner of the Google Solution Challenge 2023.
Overview
BuzzBusters applies computer vision and cloud infrastructure to support the detection and monitoring of Aedes aegypti mosquitoes, the vector of dengue and other diseases.
Problem
Prevention campaigns depend on knowing where the mosquito is present. Manual identification is slow, requires expertise, and the resulting data is rarely centralized or geolocated.
My role
Co-founder, AI and software engineer. I worked on the computer vision model, the cloud architecture and the product experience together with my team.
Solution
A mobile application that lets people capture an image, receive an AI-assisted identification, and contribute the report to a geospatial dataset that can inform prevention work.
Technical approach
Computer vision models trained with TensorFlow and served through Google Cloud and Vertex AI, a Flutter client, Firebase for application services, and BigQuery for analysis of geospatial reports.
Challenges
Working with limited and imbalanced image data, keeping inference usable on everyday mobile devices, and designing a reporting flow simple enough for non-technical users.
Results
The project was selected as a Global Winner of the Google Solution Challenge 2023, reaching the top of a worldwide student solution competition.
What I learned
That applied AI creates value only when it is embedded in a workflow real people can use, and that health problems require listening to the institutions that will use the data.
Technologies
- Computer Vision
- TensorFlow
- Google Cloud
- Vertex AI
- BigQuery
- Firebase
- Flutter
- Geospatial data