Founder and Data Scientist
2014 - Present (over 5 years)
Client projects included: • Predicting patient no-shows for a large health care provide with a distributed machine learning model • Optimizing voter mobilization for...more a congressional campaign, using python for modeling and d3.js for visualization • Developing a campaign optimization system for a political group using TensorFlow, Hadoop with mongoDB for storage and ETL, and serving predictions via a service built in node and express, with a simple react app for querying and displaying results.
Sr Data Scientist operating across the company's competencies, with a particular focus on KGI's Healthcare and Life Sciences practice. Recent client engagements include: •...more Building a web application using Spark, R, and d3.js that processed terabytes of data to predict likelihood of clinical trial success • Creating a system for AI assisted planning and design of clinical trials at a cutting-edge Life Sciences company • Developing a pricing engine for the TTS division of a major bank
Led the successful creation and development of HMS’s data science department, with technical responsibilities including: • Replacing legacy solution for identifying fraudulent...more healthcare claims with a deep neural network built with Tensorflow on Spark • Leading company-wide effort to adopt data science best practices and rapidly move from outdated legacy technologies
Consulted with Zynga's Advanced Technologies Group on several projects including: • Creating a React Native augmented reality mobile app prototype with a node.js and mongoDB...more backend • Developing backend services to rapidly process big data using node.js
• Contributed to development of automated data pipeline for computation of advanced analytics, built with Hadoop • Implemented deep learning network in python using Theano...more for GPU parallelization to predict individual-level annual healthcare expenditures
Common Sense Action
2013 - 2014 (7 months)
• Supervised the development & implementation of a new data-driven outreach & mobilization strategy for this Millennial-focused political organization • Trained...more a deep learning system to predict first-time voter behavior in a probabilistic fashion • Used python for modeling and data munging, express.js and node.js for building micro-services
Michael R Hollis Leadership Fellow
2013 (3 months)
• Led quantitative aspects in the development of a data-driven healthcare delivery model specifically designed for Atlanta's homeless population • Built delivery optimization...more models using python and MySQL
Data Scientist · This project was part of a Data Science competition held at Emory University in 2013, with the objective of extracting novel insights… · More from a massive collection of energy consumption several years' worth of device-specific kilowatt usage measured in 60 second increments. Using a GBM our model scored highest in terms of predictive accuracy, giving us First Prize Overall. In addition, we took home an award for best data visualization, which consisted of a simulated real-time dashboard showing model performance over time and by appliance.
Data Scientist · Avito.ru competition: Text analysis, NLP, content classification; top 10% finish
Liberty Mutual Group: Rare events, risk… · More analysis, highly imbalanced data; top 25% finish
I created a novel machine learning algorithm to predict individual-level voting behavior with 95% accuracy, which I deployed via a REST API and designed a minimalist web app to deliver the model's predictions. Before that, I built models for a investment firm, including similar full-stack deployment of a model that predicted successful long-term sector rotation strategies that led to a 60% increase in accuracy over the firm's legacy tool.
I would like to transition away from data science consulting in the healthcare space to working in the industry, preferably with a team from whom I can learn while taking on problems that are challenging in both technical as well as human aspects.