Building data and machine learning services to detect and mitigate fraud across all Uber’s products including Uber Rides and Uber Eats globally.
•Optimized and rebuilt a data ingestion pipeline from Python and MySql into a Java service with a Cassandra datastore. These features are used in Machine learning models and rules to detect and protect against fraudulent activity.
•Created Python pipelines with hive queries to disperse data from HDFS to Cassandra. This data was then used to create various dispersal and aggregation features exposed to our rules engine which caught 6k daily fraudulent drivers across Europe, the Middle East and Africa.
•Worked with analysts to implement features in our Golang RPC service, blocking drivers who abused airport and uber pool trips.
•Investigated and fixed MySQL rules to enable emerging products like Uber Medic to make trips to Hospitals during COVID-19.