End-to-end time series forecasting portfolio with 8 real-world applications. Implements LSTM, Transformer, CNN, Autoencoder architectures for fraud detection, sensor monitoring, activity recognition & energy prediction. Complete with preprocessing pipelines, model deployment & comprehensive READMEs. PyTorch/TensorFlow.

Trains a model with Tensorflow / Keras to detect digits from my electricity meter. Images are coming from an ESP32 camera webserver. It predicts the digits, trains itself with the new predicted images and writes the information to an Influx database. Counter status is then visualized with Grafana. Alerts are also managed by Grafana.

EnergiNet Server is a Node.js and Express-based backend for managing solar energy monitoring, forecasting, and blockchain energy trading. It connects to MongoDB for data storage, uses CORS for secure API communication, and integrates AI models via TensorFlow.js for real-time solar energy prediction and analytics.

EcoFlow AI uses practical generative AI and ML to optimize cement plant operations. Features include real-time monitoring, energy optimization, predictive maintenance via NLP, and waste management with computer vision. Built with Python, Flask, React, TensorFlow, and Google Cloud APIs.

An ML-powered smart building system using ESP32 and TensorFlow Lite. Integrates DHT11, MQ135, PIR, and TSL2561 sensors for real-time monitoring of temperature, air quality, motion, and light. Executes autonomous control of fans, LEDs, and lighting via an embedded neural network for optimized energy efficiency and occupant comfort.

AI‑powered, open‑source platform for real‑time energy monitoring, predictive analytics, and community‑driven sustainability built with React, TypeScript, Vite & TensorFlow.js.

WattWise Energy Tracker is an intelligent web application that enables real-time monitoring and 7-day forecasting of electricity consumption using AI models (LSTM, BiLSTM, CNN). Built with React, FastAPI, and TensorFlow.

A machine learning–based project using CNN to detect and predict faults in solar panels from current and voltage data. Built with Python, TensorFlow, and a web interface, it improves solar efficiency, reduces maintenance costs, and enables real-time monitoring for sustainable energy management.

This AI prototype transforms LSTM energy forecasting into a deployable smart grid maintenance system. Built with Flask, TensorFlow, and PostgreSQL, it handles load prediction, anomaly detection, and risk scoring. It bridges the gap between AI research and production-ready solutions through integrated monitoring and recommendation engines.

Developed an AI- and IoT-based smart city solution for traffic, air quality, and energy optimization. Built YOLO-based vehicle detection for adaptive signal control, implemented a TensorFlow/Keras model for AQI prediction, designed sensor-based smart lighting, and integrated IoT data with cloud platforms for real-time monitoring and analytics.