Costantinos Constantinou

Senior Data Scientist @National Gas

Warwick, GB

MLOps, Neural Networks, Azure DevOps, Docker, SQL, AI, Google Cloud Platform• Model Optimization: Improved forecasting accuracy by 23% for interconnector gas demand and 12% for wind generation models through targeted feature engineering and model stacking. Minimised forecast error and achieved incentive target by Ofgem.• Time Series Forecasting: Designed, deployed and maintained a suite of demand forecasting models, using Neural Networks, Gradient Boosting, Regression for predictive analytics. • Probabilistic Forecasting: Productionised probabilistic forecasting models using Generalised Additive Models based on ensemble weather data. Integrated copula functions to quantify uncertainty, enabling control center stakeholders to optimize bidding in energy markets. • ML Model Development & Deployment: Developed and deployed ML models to production using Azure ML and Python SDK, automating validation and delivery through CI/CD pipelines.• Azure DevOps & ETL Pipelines: Designed and automated end-to-end data pipelines for ingestion, transformation, and validation using Azure ML Studio and Databricks.• API Clients: Built Python packages for API client frameworks in Python, implementing exponential backoff, and retry logic. Published and maintained packages within Azure DevOps, improving reliability and modularity of ETL pipelines.• AI Initiative with Google’s Gemini: Evaluated and guided the implementation of generative AI to extract insights from engineering reports, focusing on project risk indicators such as cost overruns and penalties.

University of Leeds

PGCert, Artificial Intelligence

The University of Manchester

Master of Science (MSc), Electrical Power Systems Engineering

Cyprus University of Technology

Bachelor’s Degree, Electrical Engineering