# Gem Finder

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [James Raber](https://cerebralvalley.ai/u/Refresh)
- **GitHub:** https://github.com/refresh2k/gemfinder
- **Demo video:** https://youtu.be/oZZ8nfgzwYs
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/371

Gem Finder is a phone app that scans a gravel bed and flags the likelys, so rockhounds pick up the good ones first instead of eyeballing every stone.

Live phone flow, not a mock: a sidewalk scan returns 0 candidates at 85% or above with a splash-and-reshoot prompt, then a real gravel bed returns a ranked list at 85% or above via SAM 3.1 segmentation and Muse Spark multimodal anomaly ranking through the Meta Model API. Tap to zoom, then close-up: clean, focus, scratch test (fingernail, coin, glass, steel), then up to three possible identities, not confirmed, with supporting and contradicting evidence and a next check. Saves stay on device.

Gem Finder doesn't name a species from a bed photo — it ranks candidates worth picking up, then helps you check the one in your hand.

## More from Meta Global AI Developer Hackathon

- [JCZ](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/368)
- [Voizgger](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/369)
- [ClaimLens](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/370)
- [Cyphorto](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/372)
- [Haileydayz](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/373)
- [EcoNest Systems](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/374)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/371. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt