This project is part of a larger company strategy to centralize computing and transform desktop computers into simple terminals, with the aim of minimizing maintenance cost. In this strategy, all desktop computers run their OS not from a hard drive but directly from a USB thumb drives. This project, which is an ongoing endeavor, consists of a completely in-house developed software building system for the USB drives, giving a more fine grained control over the setup of the resulting Linux OS then most other solutions available. This makes it possible to develop highly tailored Linux solutions matching precise requirements for specific clients. Developed a deep understanding of the internals of Debian based Linux (live) distributions during work on this project.

Topazis is a VoIP PBX solution for SME environments, based on the Asterisk open source software. AcXi is a the desktop application that interacts with a Topazis server. This job required a diverse set of skills, including protocol and application design, software and service integration and making strategic decisions regarding system stability, security and scalability. The resulting product is a fully integrated application stack running on a Debian Linux distribution. Additionally a software build environment was made to ensure this project's life cycle and guaranty. Several features that set this product apart from others like it are: CTI (coupling of PBX with Business Information Systems), many ways intelligent call routing and Call Data Records that actually reflect reality (instead of those produced by Asterisk). The relatively new field of VoIP for SME environments is full of pitfalls and gotchas, both technical and business-wise. This project has honed a thorough understanding of the theoretical and pragmatical challenges and their solutions.

DataNavigator is a framework for making Business Information Systems. This job consisted of writing a middle-ware layer between an existing graphing library called JpGraph on one side and graph data structures produced by DataNavigator on the other side. In essence this project was primarily about finding a smart way of re-interfacing the graphing library in a way that exposed its features while at the same time giving DataNavigator an easy interface to work with. The biggest challenge of this project was finding the right middle ground between the formatting of the input data, the features of the library to be exposed and the amount of complexity and flexibility in gluing these two together.

Similar to the above mentioned Graphs module project, this project was also an extension of the DataNavigator framework. This module gives DataNavigator a uniform interface for generating all kinds of reports, e.g. invoices, inventory overview or bookkeeping metrics. The module was build around an Open Source PDF generating library called TCPDF. Together with a configuration file specifying a reports general structure and markup, the module transforms the input data into PDF that can subsequently be used for download, email or automatic archiving purposes. As with the graphs module, this project's biggest challenge was designing the proper information structures to describe reports that would give maximum flexibility without over-complicating the interface. Additionally this module also includes solutions to problems that can be found in many alternative PDF solutions, like e.g. improper page breaks and bad support for multi-line table entries.

This project was part of a customization job for a client that needed their live and legacy email scanned and subsequently processed into their BIS. Because all affordable third party and Open Source solutions available for PHP at that time appeared only to be able handle simple and perfectly formatted emails (which is not a much of a real-life scenario) a decision was made to write our own. The result of this project is a PHP CLI application that can parse an email, including attachment, into an object oriented data structure. Contrary to most alternative solutions, this program is as forgiving to format errors as possible to maximize the information that can be extracted.

This project was a ?carte blanche? design and implementation for a client who wanted a web site that she could (partly) maintain herself. The client was not very technical savvy, therefore a classical CMS was no option. Instead the editable parts of the web site, primarily product's price lists, had to be editable in a WYSIWYG like manner directly from within the web site. Because this was a fixed price and no deadline project, it also became an experiment into solving a fundamental flaw in mist modern ajax driven sites, i.e. a lack of OOP and compartmentalization on the browser side. The result was a site that follows the Model View Controller strategy, both on the server and client side, and has a highly organized structure both in source code and during runtime. Although the web site is not that special in it's content, the technology behind it is a top-grade recipe for solving code management and runtime issues with modern web sites.

A straightforward project in which a client had hired a design firm for their new corporate look-and-feel. The job was to translate the graphical designs, which came as a set of bitmaps, into a functioning web site mock-up framework. Though basically a fairly simple job, what was challenging about this project was finding a solution for some key aspects of web sites the design firm apparently was clueless about, like dynamic page sizes and limited support for exotic font types.

This project was a custom job for a client with a complex email infrastructure. It comprised of a Lotus Notes email server running on Windows and a BIS running on BSD UNIX. The tricky part here was that the BIS needed to scan all emails to and from the server in addition to sending out emails itself, destined for the server, as well. Because all should happen within a single domain, traditional mail routing tricks did not apply here. A solution to this problem was found in a web facing email relay bases on Exim4 running on Linux. This server was configured in such a way that it split all email depending on it's source and destination. In addition to the standard SMTP email delivery to the Lotus Notes server, a custom SSHFS based delivery transport had to be developed for email delivery to the BIS, because this server did not have a SMTP endpoint. Also, all email was duplicated to a remote file system location for a backup purpose.

This project was for a client that received a complex calculating module from a third party. The module needed to be integrated into their BIS. The problem was that their BIS was running on Linux and the module came as a precompiled Windows DLL. A solution was found in embedding the DLL into simple XML-RPC server written in python. This server, running on a Windows server, exposed the complete DLL interface as a web service. On the other end, a small XMP-RPC client was developed in PHP and subsequently integrated into the client's BIS. This job created a nice generic blueprint for exposing any calculative feature packed into a Windows DLL to other platforms with PHP support.

This was a private project to deal with a big problem that most if not all RSS feed aggregators appear to share. That is, none of them scales well. When confronted with large amounts of RSS sources (>100), most feed readers crumble, stability wise but even more so in organization and presentation. The product of this project is a server platform that can handle a large set of RSS feeds without becoming unstable. The web interface that comes with it can be organize feed articles in many different ways like priority, tags, keywords, topics and more. It makes it possible to manage and keeping track of large amounts of information without loosing focus in the chaos of quantity. Additionally this project has been a big help as a skills practice in complex web site design with the Zend Framework.

This project was about writing a technical construction specification for complete new SCADA system of the extraction and purification network of one of the four biggest water industry parties in Holland. The SCADA system in use at that time was dangerously outdated and did no longer reflect the reality in the field. Most of the work on this job turned out to be trying to make sense of what unfolded as an absolute horror story of field equipment inventory and absence of documentation on the SCADA system and the processes it controlled. Learned very valuable lessons during this project in business politics and diplomacy and the art of communicating with people from different educational levels and disciplines.

Not a single project, but a series of small projects. These jobs consisted of update and revision work at several factories in the steel, petrochemical and pharmaceutical industry. Part of the jobs was the assessment of field situations and to update outdated documentation and drawings. The other part was making building plans and instructions for the replacement and upgrading of existing installations. Although technically not really much of a challenge, these jobs have been a great deal of training in technical drawing skills and dealing with practical industrial automation installations.

This project grew out of a private interest in embedded internet at a time that this technology was still in its infancy. Contrary to normal practice, instead of a supervised graduation project within a company, my graduation project became an independent research job into the capabilities of embedded internet technology. The result was a microcontroller platform. All of this microcontroller's I/O was directly exposed through a web service. Also, the system had several security features to prevent unauthorized access. Finally a web server with java applets demonstrated the usage of the I/O controlling interfaces.

This project was about writing software to interact with a industrial camera through a FireWire interface. The aim of the project was to develop a vision processing system for a large car manufacturer that would assist drivers in avoiding accidents. I was responsible for writing the code that would interface between the camera and the processing unit. As both the FireWire interface and the camera technology were cutting edge technology, this project involved a lot of development through trail and error. This project has been very valuable in teaching reverse engineering skills and writing device driver code.