• Effectively managed a team of up to 7 which owned 30% of Alexa’s Natural Language Understanding (NLU) customer traffic in US English (~1.5B monthly requests)• Consistently ranked in the top quartile of managers based on employee surveys in many key skill areas such as inclusion, team effectiveness, and manager satisfaction.• Launched an internal machine learning-based social media Alexa voice-of-the-customer website used by several teams• Mentored my team members, guiding them toward new roles, skills and opportunities• Partnered with multidisciplinary teams to launch new Alexa features

• Built and deployed Alexa’s production NLU models in 5 different languages• Ran offline and online experiments that supported Alexa’s product launch in Colombia• Created a widely-used Python package to efficiently root-cause the source of changed model behavior, winning best invention during a demo event and presenting at an internal Amazon conference• Promoted to Research Scientist II, October 2019

• Performed research in machine learning and NLP applied to social media on tasks such as classification, sentiment analysis, topic modelling, named entity recognition and semantic grounding• Wrote production code in the company’s Java and JavaScript codebase related to various NLP tasks• Developed an internal interactive client-monitoring web app using Shiny in R• Implemented a text sentiment classification model based on the current state-of-the-art method• Created and documented an R package to do various statistical analyses used by the research team• Monitored and improved the performance of our classification models deployed for our clients• Collaborated with account managers to generate data analyses for client reports and ad-hoc requests

• Consulting engagements with HealthTeacher, myStrength, meQuilibrium, Transcendent Endeavors• Analyzed web usage data and pre-post surveys to study mental health outcomes, explore relationships between health knowledge and behavior variables, and find leading indicators of program participation and engagement• Created an ROI calculator to estimate the economic benefit of a mental health program• Contributed to designs of studies and survey writing• Collaborated on written data analysis reports that were delivered directly to the client

• Collaborated with a small team to explore the mobile health app marketplace, develop digital health intervention programs, and improve the user experience• Managed the creation of the company’s standard reporting suite. Prepared business requirements documents (BRDs) and led the report writing implementation team• Advised on business strategies regarding web analytics, data quality standards, proprietary algorithms and customer-facing web tools• Created data dictionary for a restructured SQL database

• Built statistical models and analyzed health outcomes and participation rates for digital health coaching programs using SPSS, Excel, and IBM Cognos cross-tabulation software • Provided analysis for customer reports using various statistical methods including t- and F-tests, factor analysis, regression and non-parametric hypothesis tests • Explored relationships between biometric and psychometric variables for internal research projects • Analyzed longitudinal data to estimate group effects and time effects

• Instructed two weekly 90-minute lab sections, held office hours, and graded homework and exams for Statistics 250, an undergraduate statistics and data analysis course• Special Mention for Teaching Award, April 2012

• Created Excel documents to be used by senior-level actuaries for non-discrimination testing, executive pension valuation, etc.• Prepared client presentations using multiple data sources• Produced detailed financial and accounting valuation reports for key clients

• Selected for the “Research in Industrial Projects for Students” (RIPS) program, a highly selective undergraduate internship • Worked as part of a team of four students on behalf of Symantec Research Labs to develop an efficient and accurate method to detect text embedded in images to improve image spam filtering rates • Served as the group’s report coordinator and wrote most of the group’s 40-page final report in LaTeX • Programmed extensively in MATLAB to develop and test new approaches to our research problem