Builder of Risk Entity Watch 👀: https://www.uber.com/blog/risk-entity-watch/

Builder of RADAR 📡: https://eng.uber.com/project-radar-intelligent-early-fraud-detection/

Contributor to Uber Michelangelo: https://eng.uber.com/michelangelo-machine-learning-platform/

• Drastically improved feature engineering system including data quality controls, fast onboarding, and experiment tracking capabilities.

• Supported the implementation of the text classification models using NLP/convolutional networks

• Supported the implementation of search ranking using xgboost- lambdamart

• Rebuilt catalog indexing pipeline bringing 10x optimization of resource utilization, consistency testing, linear scalability, and performance improvements.

• Designed and implemented the processes for data quality and search results validation

• Built the AI for detecting malicious user activities using custom-designed time-series anomaly detection algorithms. Scaled to handle 4 billion events per day.

• Build Spark-based framework for hosting streaming apps featuring the following capabilities:

• Breaking change for stateful streaming apps

• Realtime reconfiguration without restart

• Time series functions for GroupBy aggregations

• Time-series Forward fill/Backward fill

• Distributed unique counts

• Built the simulation system for the user behavior using Monte-Carlo sampling methods implemented using on Akka

• Developed AI chatbot capable of producing advanced data reports based on natural language commands.

• Real-time data stream analysis and NLP using Apache Spark, Scala, Akka, Lucene, Elastic Search, Apache Storm

• Custom NLP models, which go beyond sentiment analysis

Built no-code system for managing "Dave" - a proprietary automated financial advisor.

This technology platform:

• delivered an order of magnitude improvement in terms of employee engagement in financial transactions over the industry baseline.

• helped to raise tens of millions of dollars in funding

• grow the company to billions in assets under management.