Deployment source for the FinAccess Eswatini Next.js application and FastAPI inference service.
# FinAccess Eswatini Web
The deployable FinAccess Eswatini product: a responsive Next.js interface and a FastAPI inference service containing two validated model pipelines and their model-matched SHAP explainers.
Open the live application | Interactive API documentation
## Product experience
The selected **Signal** direction presents five connected areas:
- Overview of weighted access estimates and analytical patterns
- Financial-inclusion evidence and Model 1 insights
- Mobile-money evidence and Model 2 insights
- One Financial Access Assessment that returns both predictions
- Methodology, leakage controls, evaluation, and explainability
The result hierarchy leads with a clear answer, followed by a supporting probability and five signed factors generated from the relevant SHAP explainer.
## Deployment architecture
```text
Vercel project
|-- web/ Next.js 16 frontend
`-- backend/ FastAPI service and validated model artifacts
/api/* -> FastAPI service
/* -> Next.js service
```
The split is declared in `vercel.json` using Vercel Services. Browser requests stay on one HTTPS origin, so the assessment needs no public backend URL or normal-path CORS configuration.
## Public routes
| Route | Purpose |
|---|---|
| `/` | Selected Signal product experience |
| `/concepts/ledger` | Archived Phase 11 editorial direction |
| `/concepts/open-field` | Archived Phase 11 public-interest direction |
| `/concepts/signal` | Signal comparison route |
| `/api/health` | Model, explainer, and artifact-integrity status |
| `/api/docs` | Interactive FastAPI contract |
| `/api/v1/assessment` | Combined two-model assessment endpoint |
## Repository layout
```text
web/ Next.js application, visualizations, and rendered-route tests
backend/ FastAPI schemas, service layer, API tests, and model artifacts
docs/screenshots/ Fresh screenshots captured from the live public application
vercel.json Multi-service build …