IshVenom — Offline AI Snakebite Triage Powered by Gemma 4

IshVenom is an offline-first AI snakebite triage and surveillance system built for sub-Saharan Africa using Gemma 4.

Every year, thousands of people die from snakebites in remote regions without access to reliable medical guidance, internet connectivity, or nearby antivenom centers. IshVenom was designed to help close that gap.

The platform allows users to:
• Photograph and identify snakes using Gemma 4
• Receive medically grounded first-aid guidance in local languages
• Access offline emergency assistance on low-cost Android devices
• Locate nearby antivenom centers
• Contribute anonymized snakebite intelligence for WHO and local hospitals

IshVenom operates in two modes:

Cloud Mode
• Gemma 4 26B identifies species and generates structured emergency guidance

Offline Mode
• EfficientNet TFLite classifier + fine-tuned Gemma 4 E2B LoRA running fully on-device using CPU-only inference

The system also powers a real-time surveillance dashboard that can help health organizations monitor snakebite hotspots, allocate resources, and improve emergency response systems across underserved communities.

Built for the Gemma 4 Good Hackathon by Google DeepMind × Kaggle.

Tech Stack
• Gemma 4 26B + Gemma 4 E2B
• React Native / Expo
• llama.rn
• TFLite
• Node.js + Postgres + PostGIS
• Next.js dashboard
• Offline CPU inference

GitHub:
www.github.com/calyxish/ishvenom

Powered by Gemma 4.

Видео IshVenom — Offline AI Snakebite Triage Powered by Gemma 4 канала Calyx Ish ♾