Member of the Siri Natural Language Platform team; brought NL to ondevice starting in iOS 15.

• Spearheaded performance optimization efforts, achieving optimal memory utilization and processing speed for NL models without compromising accuracy. This involved managing cross functional efforts with modeling and other Siri teams.
• Led the effort to expand our on device platform to support functionality outside of Siri increasing our impact to user experience
• Supported new hardware bring up efforts, optimizing NL Platform per hardware platform as needed
• Primarily worked with on-device development utilizing Objective-C / Swift
• Used Python/Tableau in offline environments for analysis on performance and monitoring metrics

SWE on a NLP team under Google AI Language focused on all things related to applying new NLP technologies towards solutions for Google. My work can be summarized into 2 sections.

Research

Technologies and methodologies used for confidential internal research are:

• Implementation of unsupervised methods including clustering for managing large quantities of text.
• Utilization of LSTM networks.
• Distillation of BERT models.

Infrastructure

Developed and improved many pipelines/frameworks within Google in relation to NLP data. A majority of this work involved scalable, distributed computing in C++ with a wide variety of Google’s internal tools.
This included contributing to datasets like A Challenge Set and Methods for Noun-Verb Ambiguity and others expected to be open sourced by end of 2020. 1 Contributed to the deployment of several models (including Sentiment Analysis) that are launched in production today.