EcoSync - AI-Driven Urban Sustainability Platform
EcoSync is a full-stack urban sustainability platform that ingests real-time data from IoT sensors, including air quality monitors, smart bins, energy meters, and traffic cameras, along with satellite imagery, to provide AI-driven insights across energy, waste, and green infrastructure for city governments.
The problem it solves is that cities operate 15 to 30 disconnected digital systems with no unified view, making it impossible for sustainability directors to correlate air quality, energy burden, and green space data, which prevents data-driven resource allocation.
To solve this, I built a Django REST backend with real-time IoT data ingestion and structured reporting. I implemented modular AI modules including an LSTM-Transformer for 24-hour energy demand forecasting, YOLOv8 with a Graph Neural Network for waste overflow prediction and route optimization, U-Net for urban heat island segmentation from satellite imagery, and a fine-tuned BERT NLP pipeline for citizen report classification and routing. The platform was deployed as a modular Python and TypeScript microservices architecture on Vercel.
The platform is designed to reduce municipal energy consumption by 20 to 35 percent, cut waste collection costs by 25 percent, avoid 85,000 tonnes of CO₂e per year at scale, and deliver a 5-year net present value of 8 to 12 million dollars for a mid-sized city. It also includes a citizen engagement portal enabling 30-second issue reporting and real-time resolution tracking.
My role was as the lead backend developer, where I architected the microservices, integrated all AI modules, and ensured seamless data flow between IoT devices, the AI pipeline, and the frontend dashboard.
Tech stack used: Python, Django, Django REST Framework, PostgreSQL, TypeScript, LSTM, YOLOv8, U-Net, BERT, IoT, Vercel, REST APIs, Microservices.
