Industrial strength payments

- Automated chargeback dispute process resulting in lower fees and better chargeback winrates- Refactored the team's online model scoring to run asynchronously to allow for redundancy in the chargeback risk check- Implemented a similarity search in elasticsearch to help detect fraud trends at booking time, resulting in a 12% global decrease in fraud

- Creator and maintainer of the service owning labeling and aggregation of all of Airbnb's chargeback data- Wrote 3 new "frictions" used by millions of users, which were online flows that users would need to pass if they scored highly on the chargeback model to prove authenticity of a transaction- Implemented a contextual multi arm bandit to best select a friction given a set of features for the current user

- Authored the initial client and server side implementation of an open sourced scheduler and job queue called Dynein currently used at all of Airbnb’s Java and Ruby services.- Owner of the financial reporting Kafka events used by finance to close their books every month

Biz Team