In addition to pursuing my own R&D work, I managed a team of research data scientists. Our tools enabled clients to better understand which messages resonated with their audiences. We developed models of people's behavior and opinions using a variety of statistical and machine learning methods. We scaled up those models into production-quality tools. Applying machine learning to the problem of survey inference was of particular interest. Highlights include: - Leading the data science team behind Civis's "Creative Focus" message testing product - Creating a proprietary, machine learning-based survey weighting system, which runs in production on thousands of survey responses per day

I worked on the Data Science Research & Development team, where I developed new machine learning tools and applied statistics and machine learning to a diverse range of business problems. This included: - Developing CivisML, our distributed machine learning framework - Writing scikit-learn compatible machine learning estimators, such as the model stacking estimator that I present in the video linked below: https://www.youtube.com/watch?v=3gpf1lGwecA

For my doctoral research, I developed statistical analysis software for an experiment devoted to measuring properties of the neutrino, a fundamental particle.The data for my dissertation came from the Double Chooz experiment, which was located in northern France. My main responsibilities were the development and running of Double Chooz's multivariate statistical estimation code. This software integrated the work of the entire collaboration to provide our experiment's primary result. Additionally my group at UChicago was responsible for building and installing the outer muon veto. This detector system allowed us to detect (and then remove) the interference of cosmic rays with our experiment.

I spent a year working on the Tile Calorimeter subdetector of the ATLAS experiment at the LHC at CERN (physicists love acronyms). My work was focused on the refurbishment of the TileCal electronics. I also spent a fair amount of time developing calibration and testing software.

I worked as an undergraduate research assistant for Professor Juan Collar. Among other things, I built devices to measure rates of environmental radioactive backgrounds for low-background WIMP and neutrino experiments.