Overview:

At Cash App, I led the CX Innovation team in a company-wide initiative focused on protecting customers and strengthening brand trust through the development of a real-time fraud detection and response system. The initiative combined customer experience strategy, operational transformation, and machine learning to shift fraud operations from reactive manual 3+ day review to an AI-enabled real-time detection and resolution ecosystem capable of identifying and blocking scam activity in under a second.

My Role:

I originated the concept for a real-time fraud detection and response system and led the initiative from vision through execution across Customer Experience, Risk, Engineering, Legal, and external technology partners.

Strategic Opportunity Identification

• Identified emerging customer and brand risk patterns across public support and social channels
• Defined the opportunity to transform fraud response from manual review workflows into a scalable, real-time operational system

Product Vision & Cross-Functional Leadership

• Developed the product vision and strategic roadmap for a real-time ML-enabled fraud detection and response platform
• Evangelized the initiative across ML Engineering, vendor, application, Risk, and executive stakeholder groups to secure organizational alignment and investment
• Facilitated cross-functional workshops to align priorities, roadmap sequencing, operational workflows, and customer experience requirements

Operationalization & Execution

• Partnered with Legal and Risk teams to address compliance, intellectual property, and deployment readiness considerations
• Led experimentation, operational implementation, iterative testing, and organizational rollout efforts
• Drove alignment and execution through technical ambiguity, operational constraints, and evolving fraud patterns

Approach:

I approached fraud as both a customer experience and operational intelligence challenge, designing a system that unified detection, decisioning, routing, and customer resolution into a continuous operational ecosystem.

Intelligent Detection & Decisioning

• Designed a real-time decisioning architecture capable of detecting and classifying fraud activity in approximately 0.6 seconds
• Integrated ML-driven outputs into automated routing, escalation, and customer resolution pathways

Human-in-the-Loop Optimization

• Introduced human-in-the-loop review systems to continuously refine model accuracy and support edge-case decisioning
• Established experimentation and validation frameworks focused on balancing model accuracy, speed, operational efficiency, and customer experience outcomes

Cross-Functional Iteration & Optimization

• Partnered closely with Engineering and ML teams to iterate through technical constraints, refine model performance, and continuously improve operational workflows
• Connected fraud detection systems directly into customer support operations to improve consistency, scalability, and response quality

Solution:

The resulting system transformed fraud response operations from manual review processes into a real-time AI-enabled fraud prevention ecosystem.

Core Capabilities

• Real-time fraud detection and automated decisioning (~0.6 seconds)
• Dynamic routing and escalation workflows based on fraud classification and risk severity
• Automated scam identification and bad actor blocking at scale
• Human-in-the-loop workflows for continuous model refinement and operational oversight
• Integrated customer support and operational resolution pathways

Outcomes:

• Reduced fraud detection and takedown time from approximately 3 days to ~0.6 seconds
• Achieved approximately 98–99% model accuracy in fraud detection and classification
• Blocked more than 400K bad actor profiles within the first year
• Freed more than $2M annually (~3,850+ work hours/year) in operational capacity
• Filed 1 U.S. patent-pending innovation
• Improved customer trust through faster, more consistent fraud prevention and customer response experiences