User targeting for multi-channel marketing. Used ALS recommender system implemented in PySpark. Led to increases in sales cycle speed and in understanding of driving factors for using the product. Worked cross functionally with data, marketing and sales teams to coordinate feedback loop.
Analyzed customer behavior in webapp, including metric dashboarding, activity segmentation & clustering, chat text mining, customer recommendation system, and A/B tested feature implementations. Visualized results using Flask and plotly.
Text mined recorded calls of sales team. Used IBM’s speech to text websocket to transcribe calls, TextBlob to analyze text to identify conversation topics leading to positive outcomes, and potential fraud detection.
Collaborated with others on data team, performing ad hoc analyses, marketing visualizations, corporate risk underwriting, data enrichment, entity resolution.
Prioritized multiple tasks and remained detail oriented while under pressure.
Worked as a team with other advisors to ensure quality and speed of responses to learners.
2015 - 2016
Kansas State University
Mathematics - Algebraic Geometry
Research Excellence Award (Surowski Memorial Fellowship)
Academic Excellence Award
2009 - 2015
Mathematics major, Physics minor
2005 - 2009
Detecting seizures with Topological Data Analysis
Using Twitter’s Breakout Detection Package: Presented on the mathematics behind the breakout detection package developed at Twitter and how we used it to profile and segment financial time series data.
Twitter stream sentiment analysis
Twitter streaming sentiment analysis. Built streaming architecture for streaming tweets using NodeJS, PipelineDB, and visualizing sentiment over time, top retweets, and top words dynamically in a webapp
with express.js, plotly and bootstrap.
Data Science KC Meetup
Persistent Homology and Topological Data Analysis With Applications and Examples in Time Series Analysis: Presented on basics of persistent homology, including definitions and examples. Described multiple applications to time series analysis using persistent homology and deep learning techniques.
PyData Chicago 2016
Ongoing project to detect seizures using TDA and convolutional neural nets. Data is from CHB-MIT EEG database. Used PySpark and Dionysus to compute Betti numbers for rolling windows of EEG signals. Convolutional neural nets will be used to predict seizure presence.