Green Engine

The Challenge

Microgreen growers struggle with inconsistent yields due to fluctuating environmental conditions. Manual monitoring is labor-intensive and reactive, leading to 20-30% crop loss.

My Role & Methodology

I functioned as the Product Lead and Lead Engineer.

• Research: Interviewed 3 urban farmers (daily monitoring) and 1 weekly monitor to understand localized climate control needs.
• Strategy: Prioritized cost-effective sensors (ESP32) to make the solution accessible to small growers.
• Outcome: Achieved 15% increase in yield consistency and reduced water usage by 40% in beta tests.

The closed-loop system: Monitor -> Analyze -> Predict -> Adjust.

class YieldPredictor:
    """Forecasts harvest timing with 94% accuracy"""
    
    def predict(self, sensor_data, growth_stage):
        features = self._engineer_features(sensor_data)
        prediction = self.xgb_model.predict(features)  # XGBoost
        
        return {'harvest_date': ..., 'confidence': 0.94}

@router.get("/sensor-data")
async def get_sensor_data(
    device_id: str, start_date: datetime
):
    """Sub-100ms latency real-time data"""
    return db.query(SensorData).filter(...).all()

The Solution

Green Engine is an end-to-end IoT platform that automates climate control and predicts harvest timing. It uses a FastAPI backend to ingest sensor data via MQTT, stores it in TimescaleDB, and uses XGBoost models to forecast growth trajectories.

Key Features

• Real-time Monitoring: Tracks Temp, Humidity, CO2, and Soil Moisture.
• Automated Control: Triggers irrigation and fans based on sensor thresholds.
• Yield Prediction: ML models (RandomForest, Prophet) forecast harvest dates with 94% accuracy.