jameskoero/nyando-flood-ai: v1.0.0 — First Production Release

🌊 Nyando Flood AI v1.0.0 — First Production Release

Released: May 19, 2026 · 14:16 EAT · Kisumu, Kenya

Nyando Flood AI v1.0.0 is the first public production release of the project.

This release marks the point where the system became live end-to-end with:

• a public FastAPI prediction API
• a live health endpoint
• a deployed donor dashboard
• a production-served flood risk model
• open repository documentation covering data, modelling, backend, and frontend components

This is the baseline release before the deployment hardening and repository cleanup work that followed in v1.1.0.

✅ Live Deployments

| Service | URL | Status | |---|---|---| | Prediction API | nyando-flood-api.onrender.c… | ✅ Live | | Health Check | nyando-flood-api.onrender.c… | ✅ model_loaded: true | | Donor Dashboard | nyando-flood-ai.vercel.app | ✅ Live |

📦 Included in v1.0.0

• Production API for flood-risk inference in the Nyando Basin
• Live model serving with version 1.0.0
• Public dashboard for donor and stakeholder access
• Open project structure including notebooks, training pipeline, backend, frontend, tests, and documentation
• Model reporting documented in the repository README and metrics files

📊 Model Snapshot

• Production model name: nyando_xgb_v1
• Documented model family: GradientBoostingClassifier
• AUC-ROC: 0.9717
• F1-Score: 0.9022
• CV Mean AUC: 0.9727
• Training points: 2,308
• Training date: 2026-05-17

Note: the deployed API identifies the served model as nyando_xgb_v1, while the repository documentation describes the evaluated model family and metrics in more detail.

🧪 Confirmed Health Response

{
  "status": "ok",
  "model": "nyando_xgb_v1",
  "model_loaded": true,
  "model_path": "/app/backend/models/nyando_xgb_v1.pkl",
  "version": "1.0.0"
}

🌍 Scope

Nyando Flood AI is an open-source flood risk prediction system for the Nyando River Basin, Kisumu County, Kenya, designed to support early warning, stakeholder communication, and future resilience planning.

This first release delivers the initial live public system and establishes the foundation for later hardening and expansion.

Built in Kisumu, Kenya for flood-prone communities in the Nyando Basin.