Résumé

I’m a software architect and engineer with 20+ years of experience building web-based applications, REST APIs, and scalable, distributed back-end services. I’ve also done quite a bit of front-end, SQL, and DevOps work, though I prefer to focus on the back end.

I’ve written production systems in Go, Clojure, Rails, Django, ASP.NET and a number of other technologies. On most projects, I have been the requirements gatherer, architect, implementer, documenter, and maintainer.

I currently work for the Academic Preservation Trust at the University of Virginia, where I build digital preservation repositories, working mostly in Go.

Projects and Experience

APTrust Digital Repository

2014 - Present | University of Virginia - Charlottesville, VA

The Academic Preservation Trust (APTrust) is a distributed digital repository serving a consortium of 18 universities across the US. Depositors send high-value digital materials (audio, video, research data, photos, master scans of rare books, etc.) to APTrust, which stores them in AWS and/or Wasabi storage on the east and west coasts of the US to ensure they remain available even after regional disasters.

APTrust provides a number of services on top of AWS’s service offerings, including regular fixity checks, access control, strictly controlled deletion workflows and a searchable metadata registry. The system consists of a coordinated suite of distributed microservices written in Go, and a web application to provide web and REST access to the registry.

Technologies: Go, NSQ, Postgres, S3, Glacier, Wasabi, Docker, Fargate, Redis, Swagger/OpenAPI

Code: Preservation Services | Registry

Documentation: APTrust User Guide | Registry Member API | Preservation Services | Registry

DART: Digital Archivist’s Resource Tool

2018 - Present | University of Virginia - Charlottesville, VA

Many depositors were slow to use APTrust, DPN, and other distributed repositories, even when they had a strong need to preserve materials. One of the roadblocks was a lack of technical skill or developer time required to package materials in archival format.

DART provides a simple drag-and-drop UI, along with a suite of scriptable command-line tools to package and upload materials into remote repositories. New users were able to deposit materials into APTrust with no training. DART is built on a plugin architecture that allows developers to write modules that support new package formats, network protocols and remote repository integrations.

Technologies: Node.js, Electron, HTML5, CSS

Code: DART

Docs: User Guide | API

DPN

2014 - 2018 | University of Virginia - Charlottesville, VA

The Digital Preservation Network (DPN) was a preservation service similar to APTrust, serving a consortium of sixty or so universities. APTrust provided ingest, replication, and storage services for DPN. I designed and built the ingest and replication services using Go, and I worked with developers from Stanford, the University of Michigan, Chronopolis, and UT Austin to design and build the Rails application that served as the DPN registry.

Technologies: Go, NSQ, Ruby, Rails, Postgres, Nginx, S3, Glacier

Code: Exchange | DPN

Online Project and Resource Management

2016 - 2017 | Rivanna Software - Charlottesville, VA

Girl Friday Productions in Seattle had been using an enormous Excel spreadsheet to manage hundreds of publishing projects and dozens of employees and contractors. After working out business rules and use cases, I built a web-based application to help the company manage assignments, personnel, deadlines, billing and reporting.

Technologies: Python, Django, Apache, Postgres

HTML5 Online/Offline Interactive Catalog

2013 | WillowTree Apps - Charlottesville, VA

A supplier of building materials needed to put its entire catalog into an offline, tablet-based HTML5 application, so that builders touring large construction sites that have no WiFi can order electrical fixtures, plumbing fixtures and other items as they walk from room to room. The application would send the orders to the supplier later, when a connection was available.

The primary challenge in designing this application was to model the correct data schema. The architecture was not particularly complex, but the implementation pushed the limits HTML5’s capabilities on several fronts, notably data storage and performance. The fact that the engineers at WillowTree were able to build this app after I left is a testament to their skills and abilities.

Technologies: HTML5, JavaScript, CSS

Emergensee

2012 - 2013 | WillowTree Apps - Charlottesville, VA

Emergensee was a personal security app that ran on iOS and Android. It streamed audio, video, and location data to servers running on Amazon Web Services. Local police defined a geographic region they wanted to monitor through a web UI, and then were able to see all activity in the area. They could communicate with users who were live-streaming incidents through a built-in chat service.

Most of the customers for the app were universities that wanted an easy, reliable way to discover and respond to campus incidents. The university licensed the monitoring station and the students got the app for free. The app was commonly used by people who found themselves in vulnerable situations, such as women walking alone late at night, who wanted to alert campus police to their location, and let police see and hear what was going on around them.

Technologies: Python, Django, Postgres, AWS Elastic Beanstalk, S3

LendStreet Experian Integration

2012 | LendStreet.com - Oakland, CA

LendStreet helps people get loans and consolidate debt. When the company was first starting in Charlottesville, I wrote the Experian integration that enabled the company to run credit checks.

Technologies: Ruby, Rails, XML

Data Warehouse and Reporting Application

2012 | Seven Simple Machines - Seattle, WA

A non-profit organization in New York City wanted to analyze years of data about its teachers and programs, and the tens of thousands of students it had served. I built an online application to warehouse data in a format that allowed for flexible querying and reporting. The system pulled in current live data on request, and included a point-and-click query builder, saved reports, and a number of other features.

Technologies: ASP.NET MVC, C#, SQL Server

Roomkey.com

2011 - 2012 | Hotelicopter/Roomkey.com - Charlottesville, VA

I built Roomkey’s initial rates and availability engine, which communicated with the back-end system of major hotel chains to find available hotel rooms based on location, date, price range, and other criteria. The system was highly scalable, often handling millions of transactions per day. Getting the system to scale required a lot of JVM profiling and some tuning.

I also wrote systems for importing data from partners and auditing the integrity of Solr and SQL databases and helped define triggers for load balancing.

Technologies: Clojure, Solr, MySQL, AWS Elastic Beanstalk

Hotelicopter.com

2010 - 2011 | Hotelicopter/Roomkey.com - Charlottesville, VA

Roomkey.com was originally Hotelicopter.com, which ran on Rails instead of the more scalable Clojure. At Hotelicopter, I designed and wrote code to extract, load, and transform large datasets from a number of hotel chains and online travel services (Expedia, Orbitz, Priceline, etc). I also wrote and maintained code that ran on AWS’s Elastic MapReduce to transform millions of images.

Technologies: Ruby, Rails, C, Solr, MySQL, AWS Elastic MapReduce

DatStat Discovery

2009 - 2010 | DatStat, Inc. - Seattle, WA

DatStat Discovery was a web-based product for researchers to design and run clinical trials of human subjects. It enabled users to design protocols for multi-arm longitudinal studies, to randomize and enroll subjects, to manage their participation, schedule appointments and assessments, and collect and analyze data. I designed and built this system after observing common patterns in a number of custom study-management systems I had written in prior years.