Hi, I'm Khaled Hasan

Challenge: Financial institutions lose billions to fraud. Traditional methods fail with severe class imbalance (frauds < 0.2% of transactions).

Solution: Developed deep learning models using PyTorch with Focal Loss for imbalance handling, ensemble methods combining Random Forest, Neural Networks, and LDA with optimal threshold tuning.

Impact: Achieved 0.95 F1-score (recall) and 0.93 F1 macro average with Neural Network. Random Forest delivered 0.93 F1-score, significantly outperforming baseline methods.

Challenge: Marketing teams needed data-driven insights on campaign effectiveness across 4 languages and multiple regions to optimize ROI.

Solution: Analyzed conversion rates, performed customer segmentation using clustering, and created visualizations identifying high-performing markets and optimization opportunities.

Impact: German market showed 0.70% conversion (3.5x average), Arabic 0.50%, enabling data-driven budget allocation saving estimated 40% in marketing spend.

Challenge: Analyze 120 years of Olympic data (1896-2016) to uncover athlete performance trends, country dominance patterns, and medal distributions.

Solution: Designed normalized database with 4 tables (Athletes, Countries, Summer/Winter Games), implemented complex SQL queries using window functions, CTEs, and multi-dimensional analysis.

Impact: Revealed top-performing nations, athlete demographic evolution, sport-specific patterns, and gender participation trends enabling historical insights and predictions.

Challenge: Hotel management needed insights on revenue trends, guest preferences, meal/room performance, and seasonal patterns for strategic decisions.

Solution: Built Power BI dashboard with star schema, advanced DAX measures for time intelligence, interactive visualizations with drill-through and dynamic filtering across 3 analytical pages.

Impact: Enabled real-time revenue tracking, identified high-margin room types, optimized meal offerings, and improved demand forecasting for capacity planning.