David Rolsky

At Thomson Reuters, I have worked on integrating new data feeds with an existing database of corporate information. Much of this work involved talking to the business side of the company, as well as the outside feed partner, and making sure that the solution I was building would satisfy our business requirements.

On the technical side, this work involved building Perl scripts for batch jobs that were run via Autosys, and updating a complex Sybase database.

At LiveText, I worked on a variety of projects. One project involved adding automatic draft saving to the FCK editor that was used to create documents in the system. As users edited, work was regularly saved to the server using Ajax. If a document was not saved "properly" by the user, perhaps because of a browser crash, they did not lose their work.

I also helped improve internal practices by implementing automated unit and integration testing. I set up a continuous test server that checked out branches of the code, ran the tests, and reported the results to a Smolder instance.

As part of a growing startup, I worked on numerous projects at Socialtext. I created our first automated deployment system, as well as our first automated tests. I also performed on-site field engineering for a major customer, integrating our appliance into their custom SSL certificate authentication system.

One of the larger projects I worked on was to move some of our data storage from the file system to a Postgres database, providing data integrity, increased query-ability, and much improved speed of data retrieval. Along the way, I also implemented a simple but flexible ACL system, and added real authorization checks, plugging numerous security holes.

One of my first consulting jobs was with Marigold Technologies, working on their salesperson-focused CRM, ContactBook. I was initially hired to do code review of the existing system. After this was completed, I helped them re-architect the code for greater scalability, and added a number of features. Along the way, I mentored another developer on the new system architecture.

The most challenging of these new features was adding per-client custom fields for contacts. The end result was customizable system for end users, while still maintaining strong data integrity constraints and allowing for quick searches over large sets of contacts.

I've also worked for Kineticode setting up the Bricolage CMS for several Kineticode customers.

Later, I worked on a simple mini-CMS for Native Americans in Philanthropy. This CMS allows them to control the dynamic portions of their web site, such as job listings and news. It is also integrated with a membership management CRM I wrote, allowing new members to sign up and pay online.

Please see my list of clients for further details.

I was responsible for the design and development of an online system "donation shopping" system. Non-profit organizations were able to enter data about themselves, and data was browseable by potential donors, who could donate directly to the non-profits.

Clients were provided co-branded access to the system, and had an administrative interface from which they could view statistics, filter non-profits shown on their version of the site, and select featured projects for their site.

In addition, I also created a development environment which allowed both the designers (using Macs) and developers (using either Windows or GNU/Linux) to share a CVS repository and to run their own copy of the site.

The site was mostly created using Mason, with a little PHP (not my work), and MySQL as the database. My data modeling tool Alzabo was used in the development of the schema for the application, as well as for communicating with the database.

I was involved in the design and implementation of a Unix version of O'Reilly's WebBoard, which had begun life as a VB web application for Windows. This was not a port in a strict sense, as the "ported" version was in Perl. In addition, many new features were added to this version which were never present in the Windows version.

The ported application used Apache and mod_perl as its platform, along with a database back end capable of using any one of four databases, MySQL, Postgres, Sybase, or Oracle.

There were a number of challenges while working on this application, including creating an installer that would seamlessly use an existing Apache/mod_perl installation or install these components from scratch. The installer also had to handle installing the software on the file system and creating the schema for the chosen DBMS. I also worked on the DBMS communication layer in order to minimize duplicated code, taking the original Sybase code and reworking it to support all four platforms. This product also supported multiple authentication models, a very full featured and customizable interface, and a powerful web interface for administration.