Usman Hasan

I've built recommender systems at Afiniti and risk analytics solutions at Bobtail, turning large-scale behavioral, financial, and interaction data into production-ready decisioning models.

At Bobtail, I designed risk features across 20,000+ monthly invoices, segmented brokers with K-Means, and reduced exposure to high-risk brokers by approximately 20%. I also streamlined Python and PostgreSQL ETL pipelines through GitHub Actions, improving downstream data accuracy from 92% to 99%.

At Afiniti, I developed scalable Python and SQL pipelines for recommendation workflows, engineered 30+ user and product features, and improved model performance by 5–10% across product recommendation tasks. My A/B experiments demonstrated 8–15% uplift in conversion and product adoption, while MLflow and Argo Workflows improved experiment reproducibility and iteration speed.

Earlier at Daraz, I analyzed millions of transactional records, built interactive dashboards, and delivered seller-performance reporting that helped operations and business teams act on trends, anomalies, and KPI performance.