Indiana University Health Physicians, Clinical Business Intelligence and Analytics* Created a Power BI dashboard for the Chief Operating Officer to answer questions such as:How many primary care patients are needed to support a specialty provider?How far do patients travel for specific procedures, virtual or in-person?How many clinical full-time equivalents are needed per specialty to serve a patient population?* Created Tableau dashboards for driving improvements in clinical operations. Identified annual savings of $5M. Gave invited talk at Vizient Clinical Practice Solution Center annual conference (clinicalpracticesolutionscenter.org, 2018).

Indiana University Health Physicians, Clinical Business Intelligence and Analytics* Created general-purpose forecasting by parallelizing the Python package SARIMAX. Forecasts were used by administrators to create annual targets for key performance indicators. Presented at internal meeting.

* Created outlier detection models to identify anomalies in site traffic. Automated generation of dashboards and daily reports for leadership team.* Served as interim data engineer; built a data service for the user dashboard featured in AdWeek:http://www.adweek.com/digital/site-young-millennials-democratizes-content-showing-full-spectrum-viewpoints-175629* Wrote Python tools for the data science team to connect to Microsoft Azure, Google Analytics, and Google BigQuery.* Designed the Python package structure, unit tests, logging, and Docker deployment process for the data science team.Technical experience: Projects, methods, and languages.* Traffic anomaly detection: Cluster analysis, feature creation, feature selection, K-means, PCA, random forest, resampling statistics. Docker, HTML, SQL, Python (jinja2, pandas, scikit-learn, seaborn, weasyprint).

* Created a JSON-based network of metrics to scale reporting tasks.* Recommended models for patient relationship management to predict patients at risk of hospital readmission. Aggregated time-series data from activity logs, call logs, and patient profile databases.* Created percentile-based metrics to assess the success rate and productivity of follow-up calls from a healthcare call center. Recommended data-driven guidelines for call center administration.* Reviewed databases for integrity. Found that performance metrics had about 2x increased uncertainty due to variable rates of data entry error. Recommended procedural fixes for data entry.Technical experience: Projects, methods, and languages.* Healthcare predictive analytics: Cluster analysis, random forest, Bayesian blocks, geocoding, natural language processing. SQL, Python (nltk, geopy, pandas, scikit-learn, seaborn).

* Built a data management pipeline at Texas Advanced Computing Center to enable collaborative data analysis (data mining demo presentation: goo.gl/usOi6n; Python use-case presentation: goo.gl/rKDqJK).* Published a Python package for stellar astrophysics, "binstarsolver": pypi.python.org/pypi/binstarsolver.* Built a time-series image analysis pipeline to automate detection of variability in star brightness. Pipeline uses machine-vision/-learning techniques to accommodate data variety (github.com/ccd-utexas/tsphot).* Led research group in adopting GitHub to speed development and promote collaboration.* Built a camera control system for UT McDonald Observatory. Interfaced camera with GPS receiver to record millisecond precise timestamps. New camera system supports larger variety of research programs.Technical experience: Projects, methods, and languages.* Time-series image analysis: Bayesian MCMC. Python (pandas, scikit-image, emcee, matplotlib).* Archival database: iRODS (PostgreSQL), Python (pandas, Google Drive API), MySQL, Bash.* Camera controller: Modified GUIs via API with Visual Studio; C, C#. Online analysis; Python (pandas).

Oversaw 100 apartments for compliance with health and safety policies. Organized community events and collaborated with local police to manage a neighborhood watch community safety program.

* Created a pipeline to detect disturbances in the ionosphere that disrupt radio communications; identified transient events in time-series data from GPS signals using wavelet transforms.* Implemented just-in-time compilers to speedup data processing by 30x.Technical experience: Projects, methods, and languages.* Time-series event detection: Wavelet transforms, NetCDF4. Python (numpy, scipy, pandas, statsmodels, numba).

* Wrote and profiled a pipeline for automating data compression within a multi-site NoSQL database. Delivered report to engineers with emphasis on maximizing storage, minimizing bandwidth, and minimizing server load.* Benchmarked Python frameworks for Hadoop. Evaluated scalability across test cluster (20 nodes, 20 cores/node) and data set sizes (1 TB). Delivered report to engineers with emphasis on minimizing time-to-solution for users.Technical experience: Projects, methods, and languages.* NoSQL data compression pipeline: iRODS (NoSQL system built on PostgreSQL), C, Python (subprocess).* Hadoop Python frameworks: Map-reduce tests of word-count, sort, K-means. Python (disco, scikit-learn), Java, Hadoop.

Taught astrophysics concepts to students at West Ward Elementary School, Killeen, TX. 20 students per lecture; 15 lectures per 2-day session; 2 sessions per year.