Aquahub: Predictive Maintenance for Biomethane

An advanced MLOps project focused on the prediction and optimization of biomethane digesters. By analyzing IoT sensor data, the system predicts chemical imbalances before they cause production downtime. View it here.

MLOps & LLM Tool-Calling

  • Predictive Modeling: Designed and trained machine learning models (using Scikit-learn and PyTorch) to forecast biogas yield based on historical sensor telemetry.
  • LLM Assistant: Integrated an LLM-powered interface with tool-calling capabilities, allowing plant managers to ask "What is the predicted yield for tomorrow?" in natural language.
  • Azure MLOps: Deployed the entire data pipeline (extraction, feature engineering, and inference) onto Azure ML and orchestrated via Azure Data Factory (ADF).