London, England, United Kingdom
Led high-impact machine learning and quantitative initiatives to enhance energy supply-demand forecasting, optimize DAM and ID trading, and deliver robust PPA solutions.
• Worked with the forecasting and trading teams to predict consumptions for the UK electricity and gas portfolio. • Joined the short-term demand forecasting project for large scale distributed ML system that offers detailed MPAN-wised power consumption profiling and forecasting; led the development on predictive AI models using time series modelling, parallel computing, Numpy, Pandas, Catboost, Xgboost Lightgbm, Pandas UDF, PySpark, Databricks, Snowflake, etc. • Automated the Excel and SQL based manual operations successfully for the forecasting team, using Integer programming, Oracle, Excel VBA, Pandas, Pypiwin32, Openpyxl, Httplib2, Scipy, Tkinter, etc. • Investigated the Unidentified Gas (UIG) with statistical modelling and random forest; assisted the team in gas market risk factor inclusion using Gaussian Copula, Hurst Exponent, etc.
• Collaborated with the R&D team and developed a real-time AI-powered proactive recommendation system, using computer vision, neural network pruning, reinforcement learning, and etc., with uses in financial risk management, in-store LED billboards and exhibition venues. Two patents were granted and one patent application is in progress. • Gained sponsorship from Imperial College Innovations to promote the Quantex PI system in the UK • Utility Model Patent "Intelligent Advertising System", Obtained, Ref. 2017201156834 • Utility Model Patent "A Pedestrian Detection System", Obtained, Ref. 2017203221823 • Invention Patent "Intelligent Advertising Methodology and System", In Process, Ref. 2017100678418