Work

Portfolio

Despite starting with limited Rails experience, contributed many new features and refactors
Researched and implemented libraries such as pundit (authorization) and paperclip (attachments) to bring new features with minimal effort
Refactored the white-labelled application from ubiquitous switch-cases to implicit path selection (e.g. via dynamic dispatch)
Fostered a software engineering mentality within the team by discouraging special cases and limiting code complexity
Served as interviewer for several usability studies with many of our users

Designed & implemented a board-game style webapp with two front-end devs and a designer over a weekend
Stored and served HMDA data (>50 million records) utilizing carefully tweaked Postgres indices
Configured and maintained demo server, a particular challenge given the amount of data we queried

Served with a brilliant team (4 devs, 2 design, product owner, PM) in 2-week sprints, with two major rollouts
Led the bureau in best practices: code commits, reviews, coverage, tests, PEP8 standards, and UI testing (Sauce Labs)
Co-wrote a Python parser for regulations (plaintext + XML) with lxml, PyParsing, python-constraint, and inflection
Derived hierarchical structure from a stream of paragraph markers using constraint programming & limited heuristics
Discovered citations (including implicit context) and definitions (including scope) via parser combinators & regexes
Automated regulation compilation (from final rules) by tokenizing, normalizing, and processing amendment texts
Created multiple output options for these regulation trees, including an API, the file system, and even a git repository
Designed and later migrating a data storage API from Flask + Elastic Search to Django + Haystack, MySql, and Solr
Helped develop a Django+Backbone-based regulation viewer, heralded as “user-friendly” and “magnificent”
Cited by the White House in the “Second Open Government National Action Plan”, December 5, 2013
Using cachegrind & jmeter, optimized many segments of a quite-intensive rendering pipeline to minimize load times
Configured and maintained demo and continuous-build environments as well as Travis-based automated test scripts
Shared code via pull requests (Github); code built using pip, virtualenv, buildout, node.js, and django-south

Designed & implemented an Idea Scale-like platform for Django on a team of 3 devs and 2 designers in two weeks
Utilized Solr's more-like-this API and stop-word configuration to reduce the number of duplicate submissions
Our code was eventually picked up for a bureau-wide rollout of the application, saving a great deal of money

Worked on a capital-Agile team: three devs, two designers, product owner, and scrum master using two-week sprints
Parsed and loaded decennial census tract shapes, population metrics, and HMDA loan applications into Postgres/GIS
Using an API-first approach, pre-rendered shares as GeoJSON and separated statistic and shape endpoints
Served data via a Leaflet-driven UI; data loaded as needed to allow for infinite exploration without high load times
Configured and maintained demo server, integrating with Jenkins for continuous deployment and Travis for testing

Successfully replaced an existing, ColdFusion under-the-desk server with a fault-tolerant, Django-based, EC2 environment
Led backend development among a team of varying size ranging from four to eight
Implemented callbacks and hand-offs with the pay.gov payment processor
Helped navigate a waterfall-driven client, agile developers, and a user-driven research team through to a project launch
Contributed to upstream projects, adding needed features and Python3 support
Focused on user security, GPG-encrypting data at rest and deleting it as quickly as possible
Tweaked our Cloud Formation setup based on feedback from load tests
Reviewed and sought review from my peers via pull requests. All changes were seen by multiple pairs of eyes
Project released to support President and First Lady Obama's Let Girls Learn initiative