Matt Jacobs

Roadblock title:

Fraud detection for developers: Unveiling the rapid model and strategy refresh framework

Time:

Thursday - 2:00 PM (5th floor - Manchester F)

Abstract:

We present a novel framework: the Rapid Model and Strategy Refresh framework. This framework is tailored to the needs of fintech companies facing escalating fraud trends amidst rapid business growth. By leveraging automation and advanced analytics, this framework streamlines the optimization, evaluation, and online adjustment of domain-specific models. Unlike the conventional periodic updates, our approach enables organizations to refresh models and strategies with agility, reducing Time-to-Market to a matter of days rather than months. In this talk, we delve into the design and implementation of the Rapid Model and Strategy Refresh framework, highlighting its key components and benefits. We demonstrate how this framework empowers developers to preserve trust in the payment ecosystem.

Bio:

Matt Jacobs is a Machine Learning Engineer with a background in physics and over three years of experience in MLOps and automation at PayPal. He has spearheaded several key initiatives, including the Rapid Model Refresh framework for PayPal’s offline modeling solutions and the optimization of distributed TensorFlow model training for large-scale data scenarios. Matt has also led efforts in automating and optimizing feature selection methods, pioneering the integration of Explainable AI (XAI)-based feature selection at PayPal. Additionally, he developed the company’s first in-house generative AI coding assistant, designed to support machine learning scientists and engineers in their coding tasks.