📹 [Webinar] Building a Real-Time Fraud Detection System at Signifyd

Fraudsters are always evolving their tactics, such as using AI and LLMs, to bypass detection. 

To combat fraud, Signifyd, an e-commerce fraud detection platform, uses ML to make instantaneous decisions on whether transactions — such as returns, chargebacks, and customer account creation — are legitimate.

Nathan Ackerman, SVP of Engineering & CISO at Signifyd, joined Tecton CTO Kevin Stumpf in a recent discussion to dive into the details of Signifyd’s platform, including: 

• The challenges they face iterating on a system that needs to continuously adjust to new types of fraud, such as scammers using LLMs to bypass gibberish detection

• How having more historical data and transactions they can use as reference improves their fraud detection capabilities — for example, Signifyd uses data from 600+ million global wallets to inform whether a transaction happening at any moment is legitimate or fraudulent

• How feature freshness managed by their feature platform allows the Signifyd team to focus on understanding trends and disrupting fraud rings, as well as enable increasingly faster fraud response