Blue Tiger is a digital consulting firm focused on designing organizations, building software, and crafting experiences that improve the lives of the American public.

I lead the data and technical architecture group for 18F consulting.

Worked on openFOIA, a project that aims to improve the Freedom of Information Act request experience. I fulfill two roles on this team: project management, and back-end software development. As a project manager, I have helped focus and lead the openFOIA team by introducing an agile process to the product development. As a software developer, I have helped design and develop much of openFOIA’s backend. Technologies used: Python, Django, Postgres. As co-leader of 18F’s Agile Guild, I have co-authored an agile guide (https://pages.18f.gov/agile/) and presented it’s contents to a government wide audience of project management professionals.As project manager for Every Kid in a Park, I have lead the team to produce a kids friendly website for 4th graders in a period of a couple months. My roles have also included producing the original wireframes for the website, driving the product direction, functioning as the sole back-end developer and wrangling the content.

* Designed and developed a web-based application that makes regulations easily readable, accessible and comprehensible. As part of a team of two backend developers, I have made significant contributions to the following efforts: (i) architecting the application and defining the data formats, (ii) developing the parser (turns regulation text into data) and (iii) developing the Django application that displays the regulations. Technologies used: Python, Django. * Developed an application to collect ideas throughout an organization, allow comments and voting on ideas. This application was developed in two weeks. Technologies used: Python, Django, MySQL. * As a personal project, I designed and developed a tool to help gist online news articles about a provided topic. The tool uses natural language processing techniques to find and display significant commonly co-occurring words, as well as named entities (people, places, etc.) giving users a way to identify important topics in the news daily, without reading the articles. Technologies used: Python, Django, NLTK

* Designed, developed and deployed all of the back-end code for http://crowdbreaks.com, a disease surveillance system based on open-source social media data. Crowdbreaks collects and processes a million tweets per day that match a set of disease-related keywords and aims to provide early warnings of emerging health issues in a community. Tweets are classified as relevant based on a supervised naive Bayes classification algorithm and reverse-geocoded based on GeoDjango and TIGER/Line shapefiles. Technologies used: Python, GeoDjango, NLTK, PostgreSQL, EC2, S3* Created an online application to survey high school students on health profile and health-related behaviors, as well as gather contact network information. The software allows one to run multiple survey campaigns. Technologies used: Python, Django * Developed a machine learning sentiment analysis prediction framework for tweets. This was used to survey public opinion on a novel vaccine. Technologies used: Python, NLTK, MySQL, Django.

* Designed and implemented a version history system that tracks markup, metadata, and file changes to documents in DocFinity. Such histories are used as audit trails, and to understand the evolution of documents. * Added the ability to print and fax multiple documents, directly from DocFinity's web interface. The system can be set up so that large print and fax jobs can be sent to, and processed on, non use-facing DocFinity servers. * Actively involved in debugging and solving problems with the DocFinity application under heavy load. The challenges tacked have involved investigating database deadlocks, high application latency, and determining database connection pool and scheduler thread pool sizes.

Devised a user feedback-based system for the main website to provide users with popular links based on their search terms. The system logs keywords and the resultant links that are clicked on. Learning from this user behavior, the system returns the more popular results at the top of the page. Technologies used: PHP, MySQL, CSS, HTML and JavaScript.

Member of a five person team that maintained and developed web-based policy and claims administration software for the Florida homeowner's insurance industry. Technologies used: Java, JSP, Oracle and IBM Websphere.

* Developed a web portal to high performance visualization services. * Rapidly prototyped software to improve user interaction with mechanical theorem provers.