Featured Projects & Research

Competed in the SPE DSEATS Africa Region Datathon 2026, engineering a predictive machine learning pipeline to forecast the presence of oil in unexplored locations using complex geological and geophysical features (porosity, permeability, trap types, and seismic scores).

Designed and executed a comparative study between two approaches: a purely data-driven ML pipeline and a physics-informed pipeline that leveraged petroleum engineering domain knowledge to detect and correct geological anomalies. Built and optimized a final soft-voting ensemble model combining Random Forest, Gradient Boosting, and XGBoost classifiers, demonstrating how integrating upstream domain expertise directly improves data quality and predictive performance.