Software Engineering Intern — PayPal
Summer internship on risk and operations tooling for PayPal/Venmo Mastercard flows. The mandate was simple and hard: turn multi-day complaint investigation into near-real-time triage by consolidating fraud and transaction signals.
Key Achievements
- Reduced complaint-resolution turnaround from 24-48 hours to near real time.
- Built fraud-detection workflows for PayPal/Venmo Mastercard operations.
- Consolidated signals from multiple transaction and data pipelines to support faster risk investigation.
Context
Card and wallet complaint handling often spans multiple systems: authorization logs, merchant data, account history, and risk signals. When those live in separate pipelines, investigators lose hours stitching context before they can act. That latency shows up as customer pain and operational cost.
What I Built
Fraud-detection and investigation workflows in Java/SQL that pulled together signals from multiple transaction and data pipelines. The practical outcome was cutting complaint-resolution turnaround from roughly 24–48 hours toward near real time for the PayPal/Venmo Mastercard operations use case I supported.
Why It Mattered
The number that moved was not a system metric. It was how long a customer waited while an investigator assembled context by hand — 24-48 hours of it, spent almost entirely on stitching rather than deciding. Collapsing that to near real time did not make the risk models smarter; it made the evidence arrive in time to be useful, which turned out to be the constraint that mattered.