Building a Trading Signal Pipeline for Polymarket
Four engineering decisions that separate a production-grade prediction pipeline from a backtest that only works in hindsight.
World Happiness: What 14 Years of Data Actually Shows
Using a balanced panel of 129 countries to find out what drives national happiness -- and what the numbers get wrong.
Fake Job Postings: How Machine Learning Can Protect Job Seekers
Training three ML models to detect fraudulent job postings -- and finding that the words in a posting matter far more than any metadata field.
Netflix Strategy: Why 'Older' Content Is Actually a Smart Business Decision
Analyzing Netflix's content trends to reveal two hidden strategies and how to fix the risks behind the numbers.
Moving Beyond Manual Queries
From ad-hoc SQL to a versioned, tested Modern Data Stack pipeline