🌱 PMAI - Intelligent Agroecological Monitoring Platform

PMAI is an open-source SaaS solution designed to empower small-scale agroecological producers and agricultural SMEs in Latin America. It enables real-time monitoring of environmental variables (soil moisture, temperature, luminosity), early warning alerts, and sustainability reporting, even in rural areas with limited connectivity.

🔗 Live Demo: https://pmai-saas.web.app

🚀 Key Features (MVP)

• ✅ Mobile-First Interactive Dashboard: Real-time visualization of sensor data.
• ✅ Crop & Device Management: Logical grouping of IoT nodes per cultivation zone.
• ✅ Smart Alert Engine: Email notifications based on customizable thresholds.
• ✅ Offline-First Architecture: Local data caching with background sync when connectivity is restored.
• ✅ Sustainability Reports: Exportable CSV/PDF reports for organic certification tracking.

🤖 Qwen AI Integration (Roadmap)

PMAI is actively evolving to integrate Qwen's open-source LLM capabilities to democratize precision agriculture:

1. Predictive Crop Analysis: Fine-tuning Qwen to analyze historical sensor patterns and predict irrigation needs or frost risks.
2. NLP Agricultural Assistant: A Spanish/English chatbot allowing farmers with low technical literacy to query data naturally (e.g., "¿Cómo estuvo la humedad en el Vivero ayer?").
3. Automated Executive Reports: Generating plain-language sustainability summaries from raw JSON sensor data.
4. Anomaly Detection: Identifying early signs of pests or diseases by correlating temperature, humidity, and luminosity spikes.

We plan to contribute our agricultural prompt engineering datasets and fine-tuning scripts back to the Qwen community.

🛠️ Tech Stack

Layer	Technology	Justification
Frontend	React.js, TypeScript, Vite, Tailwind CSS	Fast, modern, excellent mobile-first support.
Backend	Python, FastAPI	High-performance, asynchronous, ideal for IoT APIs.
Database	SQLite (Dev) / PostgreSQL + Supabase (Prod)	Robust, open-source, excellent time-series support.
Infrastructure	Cloudflare Pages, GitHub Actions	Zero-cost, global CDN, automated CI/CD.
Hardware	ESP32, DHT11, Soil Moisture Sensors	Low-cost, solar-compatible, Wi-Fi/Bluetooth enabled.

🏗️ Architecture

┌─────────────────────────────────────────────────────────┐
│  DEVELOPMENT (Local)                                    │
│  Frontend: React (localhost:3000)                       │
│  Backend:  FastAPI (localhost:8000)                     │
│  Database: SQLite (Lightweight, zero RAM impact)        │
└─────────────────────────────────────────────────────────┘
                          │ Git Push / CI/CD
                          ▼
┌─────────────────────────────────────────────────────────┐
│  PRODUCTION (Cloud)                                     │
│  Frontend: Cloudflare Pages → pmai-saas.dev             │
│  Backend:  Railway/Render → api.pmai-saas.dev           │
│  Database: Supabase (PostgreSQL 500MB Free Tier)        │
│  AI Layer: Qwen API (Future Phase)                      │
└─────────────────────────────────────────────────────────┘

💻 Featured Code: Smart Alert Engine

Here is a snippet of the backend logic that evaluates incoming sensor readings against user-defined thresholds to trigger alerts:

# app/services/alert_engine.py
from app.models.database import Alerta, Umbral, LecturaSensor
from app.core.config import settings

async def evaluate_alerts(lectura: LecturaSensor, umbrales: list[Umbral]) -> list[Alerta]:
    """Evaluates sensor readings against configured thresholds and generates alerts."""
    alertas = []
    
    for umbral in umbrales:
        # Dynamically get the sensor value based on the threshold variable
        valor = getattr(lectura, umbral.variable.lower(), None)
        
        if valor is not None:
            if valor < umbral.valor_minimo or valor > umbral.valor_maximo:
                severity = "Crítica" if valor < (umbral.valor_minimo * 0.8) else "Advertencia"
                
                nueva_alerta = Alerta(
                    tipo_alerta=severity,
                    mensaje=f"{umbral.variable} fuera de rango: {valor}",
                    variable_violada=umbral.variable,
                    valor_detectado=valor,
                    id_umbral=umbral.id_umbral,
                    id_lectura=lectura.id_lectura
                )
                alertas.append(nueva_alerta)
                
    return alertas

🚀 Getting Started (Local Development)

1. Clone the repository:

git clone https://github.com/appjava/pmai-mvp.git
cd pmai-mvp

2. Backend Setup:

cd backend
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload

3. Frontend Setup:

cd ../frontend
npm install
npm run dev

👨‍💻 Author & Community

Jaime Alberto Valencia Abadía Mechanical Engineer (10+ years) & Software Development Technologist (SENA). Building sustainable tech solutions, one sensor, script, and circuit at a time.

🌐 Portfolio: appjava.pages.dev

📧 Contact: java8934692@soy.sena.edu.co

🤝 Contributing

We welcome contributions from the open-source community! Whether you want to fix a bug, improve the UI, or help integrate Qwen AI, please read our Contributing Guidelines before submitting a Pull Request.