πΏ AI-powered, offline-first soil prediction system for rural African farmers. Built with React, Express, Drizzle ORM, and edge-AI simulation for the Africa Deep Tech Challenge 2025.
# πΏ NeuraSoil β AI-Powered Soil Prediction System
**By Abdulrahman Adisa Amuda**
An offline-first, AI-powered agritech platform developed for the **Africa Deep Tech Challenge 2025**. NeuraSoil empowers rural African farmers through predictive soil intelligence β even with no internet. It simulates underground water behavior, soil health, and crop suitability using edge computing and lightweight AI logic.
---
## π Overview
NeuraSoil is built with modularity, low-resource performance, and offline capability in mind. It combines a React dashboard, TypeScript backend, in-memory database, and rule-based prediction logic β simulating an end-to-end edge-AI solution for underserved farming communities.
---
## π§± System Architecture
### πΌ Frontend
- **React + Vite** β Fast dev experience
- **Tailwind CSS** β Utility-first styling with agri-based theme
- **Radix UI + shadcn/ui** β Accessible components
- **Wouter** β Lightweight routing
- **TanStack Query** β Data fetching and caching
- **Multi-language Support** β English, Yoruba, Hausa, Swahili, French
### π§ Backend
- **Express (ESM + TS)** β Modular REST API
- **Drizzle ORM + PostgreSQL** β Schema-safe data storage
- **Zod** β Runtime validation of inputs and DB operations
- **TSX + esbuild** β Hot reload dev, fast build for prod
### π§ Machine Learning Engine
- Rule-based mock prediction logic (resource-light)
- Provides:
- Soil health evaluation
- Water movement prediction
- Crop suitability detection
---
## π§ Key Features
| Feature | Description |
|---------------------------|-------------|
| π‘οΈ Sensor Simulation | Generates realistic readings every 30s (moisture, pH, pressure, temp) |
| π§ AI Predictions | Local prediction engine using rules/thresholds |
| π Trend Insights | View historic readings and predicted patterns |
| β οΈ Alert System | Triggered warnings and critical alerts |
| π Multi-language | Yoruba, Hausa, Swahili, French, English |
| π₯ CSV β¦