On my CS2 Esports Pricing Model Pipeline, I built a DuckDB-based warehouse and leakage-safe map-level training datasets. I developed a calibrated logistic-regression pricing model that reached a locked-test log loss of 0.594.
I implemented walk-forward validation, validation-only calibration selection, and model-governance controls to avoid overfitting and hindsight-driven promotion.
For my NBA Player Props ML Pipeline, I built a dataset with about 112K rows and 900+ features, and designed six-fold walk-forward validation with no look-ahead bias. I also implemented multi-layer leakage detection and synthetic sportsbook modeling.
During my IT Systems Support military service, I developed and managed databases and IT systems using MS Access, SQL, and Excel. I automated examination processes with Visual Basic and Python scripts, and published results through a WordPress CMS portal.
