salmon_boy

Salmon Boy offers not only mere data, but true wisdom
Best Use of AI Award

Inspiration

This project draws inspiration from the Pacific Indigenous legend of the Salmon Boy. A bratty contrarian, Salmon Boy disrespected the fish that formed the backbone of his culture, and wasted their precious life giving meat. After the Salmon People bring him to live in their underwater village, he learns the importance of the salmon and respecting the balance of the ecosystem. Ultimately Salmon Boy returned a visionary healer and shaman who taught his people to live in harmony with the salmon.

What it does

Salmon Boy uses satellite data to map harmful algal blooms, sediment buildup, and microplastics along salmon migration routes. By providing early warnings about these threats—responsible for the deaths of millions of fish each year—the platform helps conservationists and fisheries take action before damage occurs, supporting healthier ecosystems and more sustainable salmon runs in the spirit of the Salmon Boy.

How we built it

We built the system by first ingesting local Sentinel-like satellite data and aligning the raster bands to a common reference grid. From there, we calculated spectral proxies for chlorophyll, turbidity, and water presence, then applied water masks to focus risk scoring on relevant areas. These computations were aggregated into normalized risk scores and formatted as heatmap-ready GeoJSON grids. Throughout development, we prioritized modularity and reproducibility, implementing a 3-layer cache for efficiency, designing migration-path-aware summaries, and scaffolding endpoints for model training and future remote ingestion. By combining environmental science with geospatial computation in a Python-based FastAPI backend, we created a lightweight, end-to-end pipeline that could be run locally for rapid demos and iteration.

Challenges we ran into

Our biggest challenges included finding useful geodata on salmon migration paths in Canada required extensive searching. Processing these large amounts of data and converting them to a coordinate format usable by our program was computationally intensive and required multiple conversion attempts. We were ultimately able to build a cache in order to make use of the massive amount of information without sacrificing usability. We also had to refine the AI model repeatedly in order to stave off the encroaching hoard of bugs and ensure an efficient and user-friendly final product.

Accomplishments that we're proud of

We are most proud of developing a simple, scalable solution to a real-world problem with devastating impacts on both human and ecological well-being. By harnessing emerging technologies, we aim to support a more sustainable global ecosystem in a practical, tangible way—without requiring massive investments of money or resources.

What we learned

Working on this project taught us how powerful simple ideas can be when combined with the right technology. We learned to translate a complex environmental challenge into a practical tool. The process also showed us the value of interdisciplinary collaboration. Our different backgrounds and technological expertise required us to come together, integrating our unique strengths to tackle the project as a team.

What's next for salmon_boy

The future of Salmon Boy focuses on scaling. With additional model training and investment, even more detailed insights could be extracted from satellite data, particularly if proxies are replaced with more detailed information. For this Hackathon, we’ve focused on salmon due to the significant environmental impact of mass salmon deaths. However, the principles behind Salmon Boy are broadly applicable and could be adapted to any marine species with predictable migration routes.