GenAI for Workspace

Worked on NLP research for my senior thesis. Focused on studying conversion of Natural Language to SQL via neural networks under the supervision of Tao Yu and Prof. Dragomir Radev at Yale's LILY lab in the Department of Computer Science. Helped create the core dataset, and used PyTorch to build an augmented SQLNet model. This research was accepted to the 2018 EMNLP conference, and the publication is available online here: https://arxiv.org/abs/1809.08887.

Working Slack's Core Product analytics team. Using Presto and Hive to query big data and Python for data mining and visualization. Creating dashboards to track key metrics and extract insights, and working on projects to drive increased adoption and usage of new product features.

Worked on data science projects in Professor Vineet Kumar's group. Projects relate to Twitter network analysis, financial analyst recommendations, and the AlgoSnap project. Using R for data analysis and interfacing with the Twitter API and WRDS (Wharton Research Data Services), creating and updating MySQL databases, and creating visualizations for publications and internal use.

I hold bi-weekly office hours open to both undergraduate and graduate students for STAT 230/530: Introduction to Data Analysis. This course introduces students to the R programming language and teaches the basics of data analysis.

Used R and SQL to analyze changes in client behavior in response to new financial regulation. Built models to identify clients who present a risk to Capital Group’s 1.4 trillion dollar business. Presented work to national salesforce leaders, who will use the models for client retention.

Worked on writing statistical analysis tools for an R package for analysis of biosensor output data. Utilized variance filtering and random forests to develop feature selection tools.

Worked closely with middle school and high school students to help them build strong foundations in math and science. Worked with over 50 students.

Independently developed a novel data analysis method called Time-Spent Histogram and published research paper as primary author (http://dx.doi.org/10.1557/opl.2014.698). Streamlined tip-etching procedure for STM.