ElectricSim

SpringBootApacheKafkaTensorFlowGrafana

General Description

ElectricSim is an educational platform oriented towards the analysis, simulation, and prediction of electrical demand in the Greater Buenos Aires area (GBA). The system integrates energy, meteorological, and astronomical data sources, consolidates them through Apache Kafka, and exposes them to a deep learning model capable of anticipating variations in electrical consumption and detecting anomalous behaviors in real time.

The platform was built with a microservices architecture using Spring Boot for the backend, Apache Kafka for data streaming, TensorFlow for the ML prediction model, and Grafana for real-time dashboard visualization. It processes historical and live data to forecast energy demand patterns.

Personal Contribution

The development of the project was divided among 6 members, and I was part of the three-person team responsible for the machine learning core and data preprocessing. My role focused on designing the neural network architecture and ensuring that the raw data—ranging from weather patterns to historical consumption—was cleaned and transformed into a format the model could actually learn from.

Working on the neural network and the data pipeline taught me a great deal through trial and error. I learned that a model is only as good as the data feeding it, which led us to spend significant time refining our preprocessing logic to handle anomalies and missing values. It was a constant process of adjusting hyperparameters and testing different layers to see how the system responded to the live data stream coming from Apache Kafka.

This project showed me that building a complex predictive system from scratch is no small feat. Never having worked on a deep learning project of this scale with a large group before, I realized that collaboration is essential when balancing data integrity with model performance.