Live: https://gurehmat.github.io/Millhacks-Website/
SEED is a single-page hackathon project website I designed and built to present a climate-tech concept focused on reducing coastal plastic pollution. The site explains an end-to-end idea: using satellite spectral imagery to identify plastic-heavy shoreline zones, machine learning to prioritize cleanup targets, and drone-based systems to support collection planning.
This project was built for fast communication under hackathon constraints, with an emphasis on clear storytelling, responsive UX, and visual hierarchy.
🏆 Best UI at MillHacks — this project received the Best UI award for its design clarity, usability, and presentation quality.
- Why coastal plastic pollution is urgent.
- How spectral imagery can detect likely debris concentration.
- How a machine-learning trash-recognition approach can classify items as trash/non-trash and support prioritization.
- How a 3D prototype/model (created with Shap3r) supports concept visualization.
- How drones can support shoreline cleanup workflows.
- A clear call-to-action for demos, collawboration, and contact.
- Designed a clean, modern, mobile-first interface optimized for judges and sponsors.
- Structured content into quick, scannable sections for rapid decision-making.
- Built the full front-end with reusable SCSS component structure.
- Deployed the project as a static site for simple, reliable access.
- The trash-recognition ML algorithm was a project/team component used in the concept.
- I did not personally implement that model code.
- The ML training/inference code is not included in this repository.
- The project also included a 3D model created through Shap3r as part of the concept demo.
- HTML5
- CSS3
- SCSS (Sass) with modular partials (base, components, layout, theme)
- Vanilla JavaScript
- GitHub Pages for hosting
- Static informational website (no backend or database).
- No live satellite ingestion or real-time ML pipeline in this version.
- No production drone-control interface.
- Embed real demo footage and interaction walkthrough.
- Add lightweight data visualizations for hotspot prioritization.
- Include a concise technical appendix for model assumptions and operations.