Abdulahi Said

I've built machine learning systems and data pipelines from scratch, with a focus on data quality, validation, and rigorous evaluation.

For my MSci dissertation, I built a customer churn prediction platform that compared six classifiers on a 7,043-record telecom dataset. I implemented custom SMOTE, improved Logistic Regression recall from 72.6% to 84.1%, and corrected a data leakage issue to ensure valid model comparisons.

I also built a Python ETL pipeline for UK rail performance data, processing 1,900+ records across 25+ train operators. The project included a normalised SQL schema and a FastAPI REST API for operator performance and cancellation statistics.

Alongside software projects, I coordinate live mobility operations at London railway stations, prioritising time-sensitive assistance requests and maintaining accurate operational records. I'm seeking to apply my data and AI foundation within financial services.