From FedEx Labs to Unicorn Founder
The foundation was built at FedEx. Not in a corner office, but in the innovation labs, where Ramanand began as an engineer working on unglamorous infrastructure problems - like what happens when a package ships to a fraudulent address. He built FedEx's package intercept system: the mechanism by which the company could redirect shipments mid-transit when fraud was detected. It's a mundane-sounding achievement until you think about the operational complexity required to catch fraud at the speed of physical logistics.
He stayed at FedEx for eight years. By the end, he had founded and built its Risk Management Division for Payments and Shipping - a function that didn't exist when he joined. That pattern - spotting a gap, building the infrastructure from scratch - would repeat itself.
PayPal came next. As Head of Emerging Markets Risk, Ramanand oversaw fraud and credit risk across some of the world's most complex, underserved markets: Latin America, the Middle East, Africa. He launched PayPal's first Seller Protection programs for merchants in Brazil and Mexico - the first time those merchants had any formal protection against fraudulent buyers. He watched how the absence of trust crippled commerce in those markets. He understood, structurally, what trust does to a transaction economy.
He also understood what the existing fraud tools couldn't do. Score-based systems produced probabilities, not commitments. A merchant who acted on a good score and got burned still paid the chargeback. The vendor who sold the score walked away clean. This asymmetry bothered him. He left PayPal in 2011 to do something about it.
Signifyd launched from a two-desk coworking space in Palo Alto. The idea: use social graph analysis to match the offline identity of a buyer with their claimed online identity - cross-referencing email, IP, phone number, physical address, social networks, public records, and credit data. And crucially, back every decision with a financial guarantee. If Signifyd says the order is good, Signifyd pays if it's not.
The model required extreme confidence in the underlying machine learning - confidence that grew as Signifyd's data network expanded. More merchants, more transactions, more signal. The more orders Signifyd processed, the better the models got, and the better the guarantee held up. It was a flywheel that got harder to replicate the longer Signifyd ran.