I build decision automation tools for optimization engineers.

I ran the Decision Engineering department at Grubhub, which is a combination of software engineering, operations research, and simulation groups. These teams, which comprise over 30 people, are aligned to build software services and models that solve business problems in strategic, tactical, and real-time operational contexts throughout Grubhub Delivery. Some of the systems we directly own or collaborate on include:* Dynamic vehicle routing for restaurant delivery* Travel time and delivery estimation* Simulation of delivery algorithms and business policies* Courier scheduling models and risk analysis* Order volume forecasting

I managed the Data Science and Decision Science teams at Zoomer, prior to their being acqui-hired by Grubhub. In this capacity, I guided and developed our advanced mathematical capabilities for data analysis, business modeling, and optimization. We called our combined team Decision Engineering since we owned the analytical process from end-to-end. Our work resulted in models that were theoretically valid, engineered components which provided value to the business through direct integration into our real-time Logistics-as-a-Service (LaaS) systems.Some of the projects my team and I were responsible for included: * Dynamic vehicle routing (2.5 million orders automatically assigned, routed, and dispatched)* Near and long-term volume forecasting* Large-scale simulation of routing logistics and business processes* Real-time estimation of ground conditions, including order submission rates & service levels

Reduced company cloud computing costs by 57% by applying discrete optimization modeling techniques to server infrastructure. Rearchitected core software products, set support standards, and built hosting and software packaging infrastructure, resulting in 10x growth in enterprise software customers. Consulted for customers to create working predictive models and performant server configurations. Managed and trained a team of six software engineers.

Center for National Security / Sr. Operations Research AnalystPreprocessed data for vehicle routing problems with R, Python, and Gurobi. Built big data decision support capability against hyperspectral imagery for geospatial analysts using Python and JavaScript. Created realistic vehicular traffic simulations for large cities with Python, Go, and SUMO.Center for Advanced Aviation System Development / Sr. Simulation & Modeling EngineerSimulated and studied component and system efficiency of the National Airspace System in Java and Oracle. Designed and provided estimates for complex software models. Managed infrastructure for data acquisition from Federal Aviation Administration.

Editorial Embedded Development / Lead DeveloperManaged a team of developers in the Washington Post newsroom. Scheduled work across desks. Set programming standards and held code reviews. Supervised maintenance and stability of over 90 web applications. Mentored developers and journalists to enhance their technical skills.Editorial Tools / Web Application DeveloperBuilt award-winning data-driven journalism projects including Top Secret America, In Obama's Words, Head Count, POTUS Tracker, All-Met Sports, Money as a Weapon, and Pick Your President. Performed data parsing, coding, and analysis for investigative and national security projects using Python and PostgreSQL.

Developed ZipMetrix, a distributed US Postal Service CASS certified address standardization software. Designed, built, and deployed PoliGrid, a proprietary MapReduce-style platform for distributed computing grids using Python and C. Wrote fuzzy matching software for demographics and voter list matching. Managed relationship with USPS during CASS certification.

Managed performance of ubiquitous software Learning Management Systems. Analyzed, optimized, and re-coded poorly performing software components in Java and Perl. Analyzed and optimized Oracle and SQL Server queries.. Designed and executed weekly performance verification test of scaled models of the Learning Management System, limiting lifetime of software performance errors to within a week.