• Developed a new Ads ranking solution optimizing for deep funnel events; worked on the full end-to-end solution from custom data pipelines for model training to online serving in the complex ads delivery system
  
• Improved the Lookalike model training system by migrating to a privacy aware storage, and reducing the time spent on training by over 90%
  

Python

• Automated user profile model retraining setup process, reducing the time it takes to add a model to the retraining pipeline from weeks to hours
• Integrated the retraining pipeline with a new workflow management service to provide better failure handling and job scheduling

Python, C++, Go, gRPC, Software-Defined Networking

• Implemented network service chaining on top of Open vSwitch for the open-source project, Magma, to allow dynamic service configurations from the cloud

• Improved usage reporting accuracy of the Policy Charging and Rules Function system by implementing asynchronous session termination and flow deletion

• Designed and shipped content insertion feature for Gmail add-ons in iOS Gmail - allowing users to insert HTML content into the email using third-party add-ons during email composition

Python (Tornado), JQuery, MongoDB

• Built and maintained API and webhook integrations with shipping companies and third-party providers that processed millions of shipping updates daily
• Developed internal tools for batch modification of tracking information

Ruby (Rails), React, Go, PostgreSQL, Redis

• Integrated Appboy with the Ruby on Rails backend to track user events and trigger campaigns to send emails and push notifications
• Created a  monitoring service in Ruby for funnel metric anomaly detection

• Created a personalized study plan feature by analyzing over 7 million questions
• Developed a central login portal for all Ivy Global services utilizing OAuth 2.0