We're building a
wildlife station
from scratch.
A Raspberry Pi hidden in the backyard that listens for birds, films them, identifies the species using AI, and sends live data to a website we built ourselves.
How it works.
Wasatch backyard birds.
The species we expect to detect in our Salt Lake City backyard, based on eBird data for the Wasatch Front.
American Robin
year roundFirst to sing at dawn, last to stop at dusk.
Black-capped Chickadee
year roundCan lower its body temperature to survive cold nights.
House Finch
year roundMales get their red color from the berries they eat.
Yellow Warbler
summerOne of the first migratory warblers to arrive in spring.
Western Scrub-Jay
year roundCan remember thousands of food cache locations.
Cooper's Hawk
year roundWhen the feeder goes silent, one is probably nearby.
Four phases.
One wildlife station.
Species cards.
Generate a shareable card for any bird we detect. Download it, post it, text it to grandma.
Built on world-class research.
Dad has two degrees from Cornell — spent years walking past the Lab of Ornithology, where the research happens. Now our backyard station runs on two things Cornell built and gave away to the world: BirdNET and eBird. Full circle.
A deep neural network trained on hundreds of thousands of bird recordings. Identifies 6,000+ species from audio alone. Runs entirely on our Raspberry Pi — no internet needed for identification.
The world's largest biodiversity citizen science database. Over 1 billion bird observations. We use their API to enrich our detections with regional sighting data, photos, and species info for Utah.
Both tools are free, open, and built by researchers at Cornell University in Ithaca, NY.
Follow the build.
We're documenting every step — the wins, the bugs, the birds. Check the build log to see where we are.
Read the build log