I lead the Engineering and Product teams that build Beech Valley’s A-List platform, a combination of internal and external tools that match finance and accounting consultants with project opportunities in hours instead of weeks. Our platform powers the vetting process for the 200,000+ consultants that have applied to join our network, provides AI-based talent/opportunity matching for our 5,000+ vetted consultants, and handles the onboarding process for getting consultants up and running on their projects. We use Node, React, TypeScript, GraphQL, Postgres, Elasticsearch, and OpenAI, all deployed on AWS ECS + S3/CloudFront with the infrastructure managed in Terraform. We deploy daily through our fully automated CI/CD pipeline.

* Worked on a wide range of projects using tools like JavaScript, TypeScript, Angular, React, Node, Jest, Jasmine, Apollo, GraphQL, REST, Ruby on Rails, Postgres, Heroku, and AWS.* Frontend consultant on EdTech product whose codebase has grown from a 1,000 line AngularJS 1.2 app into a 100,000 line Angular 2+ app over 6 years. I spearheaded several large migrations that enabled the small dev team to keep up with feature requests while still writing maintainable, tested code.* As part of a team, designed and implemented a system for real-time updates and collaborative editing using Pusher, Angular, and Ruby on Rails.* Sole developer on an academic data visualization project. The client had existing designs, and I created a pixel-perfect app to match. I built the site as a static React app and used Highcharts for the visualizations.* Tech assessments for existing codebases to help clients understand the risks and opportunities with their projects. Usually, this starts because the software is getting buggier and harder to change. I help implement best practices, like automation and testing, to combat bugs and increase velocity.* Various development infrastructure improvements for a number of projects, including setting up continuous integration, dockerizing development and production environments, setting up automated deployments, and migrating build systems.

I worked on the Open Cloud Control Panel team (responsible for https://mycloud.rackspace.com). Our project was a large SPA with a Django backend. We wrote tests for all of our code (using Jasmine and Selenium), practiced continuous integration and continuous delivery, did code reviews, and followed an Agile/Kanban workflow.Aside from general feature development, I integrated Knockout.js into our existing home-grown, Backbone-inspired MVC framework (built on top of Google's Closure library) and educated the team on its use via a tech talk and code reviews. I also co-presented a tech talk to the entire Blacksburg office on continuous delivery best practices.Languages/technologies used: JavaScript (Knockout.js, Google Closure Library, Google Closure Compiler), Ruby (RSpec), Python (Django and Twisted), Selenium, Chef, Rackspace Cloud, Jenkins, Github

I created a server-centric web framework called CloudBrowser (http://cloudbrowser.cs.vt.edu/) using Node.js, CoffeeScript, and C++. The framework allowed developers to write applications using front-end technologies (JavaScript, HTML, and CSS), but the application code actually ran on the server. Clients connect to server-side instances, and a custom synchronization protocol (running over a WebSocket) keeps the client in-sync with the server and allows client-side events, like clicks, to be intercepted and processed on the server. CloudBrowser is similar in spirit to Meteor and Firebase.As part of this work, I wrote Contextify, which is a C++ plugin for Node.js that improves on Node.js's ability to run JavaScript code in an isolated sandbox (required for us to simulate the JavaScript environment of a browser on the server). As of October 2013, this library sees over 70,000 downloads a month. The Contextify code has since been merged into Node.js itself.You can find more information on the project home page: http://cloudbrowser.cs.vt.edu/Languages used: JavaScript/CoffeeScript, C++, HTML, CSS

I worked on a project where we studied software structures through a complex networks lens. I used Valgrind (with the Callgrind plugin) to collect function call data on running programs (dynamic analysis), and wrote a custom LLVM plugin in C++ to collect static relationships between functions (static analysis). I wrote a Ruby program (with a GTK+ GUI) that would parse the resulting output files and look for statistical phenomena. I also incorporated an open-source Matlab library for doing power law analysis, and parallelized it for use on a 64-processor cluster. We found that power-law relationships abound in software.A presentation by my advisor on our research can be found here: http://vimeo.com/28935715Languages used: C++, Ruby, Matlab

I analyzed run-time performance of embedded devices by hooking into the Windows CE kernel and monitoring the time spent in each thread. I used a logic analyzer hooked up to pins on the device, and my kernel code sent thread information to the logic analyzer via those pins using a simple binary protocol I came up with. I wrote a C++ program to analyze the large output files from the logic analyzer (it can collect data at microsecond resolution), and used that information to improve the performance of 2 device drivers (one by 10x, another by 3x).Languages used: C, C++