eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Rahul Pangam worked at Google for seven years focusing on reducing fraud, and now he’s taken that expertise to his startup Simility. Pangam founded Simility with the goal of reducing online fraud via a combination of human data scientists and the right amount of machine learning.
Pangam, who is also CEO of Simility, started building the company in May 2014 with a number of fellow former Google engineers. The effort has culminated in the Simility platform, which became generally available on May 26.
To date, Simility has raised $7.2 million in venture funding from a number investors, including Accel and Trinity Ventures.
“At Google, we were instrumental in building and scaling up a lot of the abuse and fraud detection efforts across various products,” Pangam told eWEEK. “At Google, we had the resources to build very sophisticated platforms, but what we realized is that the larger market had a real need for a better fraud prevention system.”
The core technologies behind Simility are a mix of open-source and purpose-built proprietary code. Pangam explained that Simility makes use of the open-source Cassandra NoSQL database to help make decisions in real time on potential instances of fraud.
“We use open-source technology as much as we can and then on top build our proprietary machine learning models,” Pangam said. “We focus on where we can make a big difference.”
The key innovation for Simility, though, isn’t its core technology but rather how it can be customized to fit specific use cases. For example, Pangam said that in an online travel business, understanding where a flight ticket is purchased and where a user is going has an impact on determining fraud. But for a gaming company, the metrics for understanding fraud are somewhat different, as it’s important to understand how a user buys an in-game credit and how soon thereafter that credit used.
“Specific attributes can make a huge difference to differentiate between fraud and regular user behavior,” he said. “We try to customize the model for specific customers.”
The challenge of customization is that it can often require professional services. Simility tries to minimize the need for professional services with a step-by-step process that drives an automated technique to enable customization, Pangam said. From a deployment perspective, organizations tie into the Simility platform by way of a line of JavaScript or an integrated API. Simility makes use of cloud data centers around the world for its back-end infrastructure.
The market for fraud detection technologies is a competitive one, including such vendors as Experian and Equifax. Key differentiators for Simility, according to Pangam, are its customization capabilities and its use of hands-on data scientists to help customers. Customers get weekly consultations with a data scientist to help continuously adjust and improve the model in an effort to reduce fraud.
“We have built a hybrid approach that lets customers have a degree of policy control and not be entirely at the mercy of machine learning models,” he said.
Sean Michael Kerner is a senior editor at eWEEK and InternetNews.com. Follow him on Twitter @TechJournalist.