xbm021

I investigate fraud, and I build the tooling that backs the investigation up.
 
For 15+ years I've worked across operations, risk, and high-volume fraud investigation. At Amazon I sat in Buyer Risk Investigations (BRI), decisioning roughly 90 to 110 cases a day at a 97% monthly accuracy rating, working payment fraud, account takeover, and buyer fraud. That seat taught me the thing that still drives how I work: a good fraud call is not just a decline. One of the saves I'm proudest of was a dormant account that tripped every velocity alarm with a sudden ~$50K order. Instead of killing it, I verified it through public professional profiles and a call to the cardholder. It was a real customer starting a real business. Catching fraud matters; not burning good customers matters just as much.
 
Since then I've paired that judgment with Python and SQL. I've built two anti-money-laundering engines: one on a 1M+ transaction dataset with seven laundering-typology detectors and a graph layer that lifted recall from ~12% to ~79%, and an iGaming/sportsbook AML simulator with a red-team/blue-team design. I evaluate my own detectors honestly against ground truth, because in this field the false-positive rate is the whole game.
 
I also founded and still run Fotorentas, a nine-year equipment-rental business where my identity-screening model held losses near 0.6% with no deposits and no credit checks. Running it solo means I've built a risk operation end to end, not just worked inside one.
 
Open to remote Fraud, Risk, AML, or Trust & Safety roles (US and LATAM time zones), as an employee or contractor.