A professional biography

I was born on June 9, 1960, in the town of Alkmaar, The Netherlands, famous for its Cheese Market. I lived in Alkmaar most of my life, until I moved to the United States. I went to school in Alkmaar, and even when I went to college at the University of Amsterdam, I commuted by train.

I went to high school at the Rijks’ Scholen Gemeenschap Noord Kennemerland (RSG) from 1972 until I graduated in the summer of 1978. During the final two years of high school, I decided I wanted to go to college, but I flipped many times between wanting to study chemistry or physics. Physics finally won out, and I started attending the University of Amsterdam in September of 1978.

I never regretted my choice and had great fun in Amsterdam. It was one particular course in the physics curriculum that had a profound effect on my career: “Computer Programming in FORTRAN.” Before I went to Amsterdam, I hardly knew what a computer was (keep in mind that most of Europe, including the Netherlands, was quite a bit behind the US in terms of general acceptance of computers).

I quickly started to like computers and programming. The FORTRAN course started us off on a big Control Data Corp mainframe (Cyber 176), where we had punched cards and a line printer as our only means of interacting with it. We also had limited access, and I could not satisfy my appetite for programming, so it was not long before I bought a TI-58 programmable calculator (about the only thing affordable for a student in those days;) personal computers were just out and too expensive). This one didn’t last me very long, and before I knew it, I had sold the TI-58 and bought a TI-59 (with more memory) and a printer. Later, I finally got access to an interactive terminal and more computer time, so life was better.

In the Dutch university system, you do get something equivalent to the bachelor’s degree (kandidaats diploma), but, unlike in the US, people generally don’t stop there; they go all the way to the master’s degree. So did I. At this point, though, you have to choose your specialization, and for me, that was High Energy Physics. The most important reason for this choice was my decision to pursue a double minor in Computer Science. Picking High Energy Physics meant I would spend most of the rest of my college career at the High Energy Physics Institute (NIKHEFH-H) in Amsterdam, where they did cool things with computers.

After completing my master’s degree in Physics in 1984, I accepted a job at the University of Amsterdam as a PhD student in computer science (in Holland, this is a full-time job, paid) all, only sometimes with teaching responsibility). I did this work at the Department of Mathematics and Computer Science: FWI. This department was a brand-new merger of the existing Mathematics faculty and the newly formed Computer Science department. For historical reasons, the latter department was an outgrowth of the very same group of people that did all the “interesting” computer work at the Institute for High Energy Physics, and a large part of the research of FWI was located in the same building and now has its own building right next to it.

My work as a PhD student involved research in parallel computing, specifically the parallel execution of the Prolog programming language. In late 1983, I first saw the Apple Macintosh and loved it. I persuaded the department to let me have one, and I took a Macintosh development course at Apple Holland. Because of my growing interest in the Macintosh and my already significant involvement in running and maintaining the department’s Unix systems (which started with a couple of VAX-750s and later grew into larger and larger networks of mainly SUN workstations), my PhD work did not progress at a normal pace. This eventually caused me to abandon the PhD work and switch jobs again. It also landed me a job as a contractor for Apple Holland, then known as Apple GEA (General European Area), where I was responsible for developer training for 1.5 years.

My next job was as the head of the computer systems group at FWI. I was now officially responsible for running all the computer systems, and I even had a team working for me to help me do it! Meanwhile, starting in early 1990, I had co-founded Storm Technology with Adriaan Ligtenberg and Chris Borton. I remained in Holland, but every weekend and night, I worked with Chris on Storm’s first software package (which eventually became PicturePress).

Links to the home pages of some of my former colleagues in Amsterdam:

Finally, in July 1991, my visa was approved, and I quit my job in Amsterdam and moved to the US (these days, I have a green card). Since then, Storm has grown from 8 to about 125 people, moved to a larger building twice, and went public (Symbol:) NASDAQ:EASY) in October 1996. I left Storm in May of 1998. In November 1998, Storm filed for bankruptcy.

Since Storm, I have started my own consulting business, called Starfield Consulting. I have alternated this consulting with jobs at startups.

I worked as CTO and Vice President of Engineering for an Internet startup, Netfreight.com (and later called Cyntric), in the area of business-to-business e-commerce in the transportation space.
I also worked as the CTO and Vice President of Engineering for CertifiedTime, Inc.
As a consultant, for about 7 years, I worked for eHarmony as an architect, and, at times, as an interim VP of Engineering. I designed and implemented front-end code (Java/Struts), optimized SQL database performance, and designed and implemented a high-availability dual-stack network architecture inside a data center. A significant other achievement was the design of a highly parallelized matching infrastructure that could scale horizontally to accommodate ever-increasing numbers of customers. I then implemented an automatic deployment infrastructure in the data center (about 120 Sun Microsystems servers) to run both front-end, database, and matching infrastructure. This was essentially what we now would call a private cloud infrastructure.

In the course of my career, I have been involved in several products:

SCSI-Power

SCSI-Power is a program that Maarten Carels and I created in 1985. We were Macintosh enthusiasts, but we were thoroughly disgusted by the poor quality and limited features of the existing SCSI drivers for the Macintosh at the time. We decided we could do a better job and started to prove it. For its time, the package was quite advanced, supporting more than 2 disk drives on a Macintosh Plus (a major achievement), partitioning, password protection, and very high performance.

We did sell the product commercially, but we never gave it our full attention, and hence it never became a commercial success, although it had, and still has, a loyal group of users. The product still works on today’s Macintoshes and PowerPCs, but is due for a major upgrade to support native PowerPC drivers and to enable asynchronous operation and “wide” SCSI transfers to achieve high performance.

During the development of all this SCSI stuff, I produced some freeware that, once installed, would significantly improve the SCSI performance of most any Macintosh Plus. The product was called SCSI-accel and can still be found in the SUMEX-AIM archives and their mirrors.

PicturePress

PicturePress was Storm’s first Macintosh application. Its main goal was to provide a means to read and save PICT files in JPEG-compressed format, and vice versa. Storm was one of the first companies to introduce JPEG software to the general market. PicturePress included fast JPEG software for the Macintosh, as well as support for Storm’s hardware JPEG accelerator.

This accelerator was a NuBus board outfitted with dual DSP16A signal processors from AT&T. This board would speed up JPEG compression or decompression about 10-15 times on a Macintosh IIfx or IIci. It allowed us to do motion-JPEG at about 15 frames per second in the same size window as Apple’s QuickTime popularized at the time (160x120).

This particular accelerator card was very successful in the market and was also sold by DayStar Digital (under the name Daystar Charger) and by SuperMac (now Radius; under the names ThunderStorm and several others).

Thunderstorm

Thunderstorm was a product that Storm developed for SuperMac. It started out as the same basic NuBus accelerator card as described above, and Storm also developed a version of the exact same architecture that would function as a daughtercard on SuperMac’s high-end graphics accelerator cards.

We also wrote new software for the DSPs on the card to accelerate critical functions in Adobe Photoshop, such as Blur, Sharpen, and Resize. It was because of efforts like these that Adobe designed a new low-level interface for acceleration into its next version of Photoshop (the Adobe Charged program).

PhotoFlash

When Storm was working on the successor to PicturePress (which was up to version 2.5 at the time), Apple learned about the product and decided to acquire the marketing rights for what became known as “Apple Photoflash”.

Photoflash 1.0 pioneered an advanced virtual memory system for images (PicturePress had always been bound by available memory) and introduced “magic” filters that automatically removed scratches and dust from scanned images. Photoflash also included an image database facility that stored references to image files (and their thumbnails) in catalogs for later reference. Photoflash 1.0 was a 68K application, but with the release of Photoflash 1.1, all major internal functions were PowerPC-native, while the UI code remained emulated on the PowerPC.

Photoflash 2.0 introduced a fully native PowerPC version and scriptability. Photoflash 2.0 was one of the first applications that was fully scriptable and recordable. It also introduced performance enhancements and several UI improvements. Photoflash 2.0 has been localized in several major languages (US English, British English, French, German, Italian, and Japanese) and is today also bundled with the Apple QuickTake digital camera.

Kid’s Studio

In early 1994, Storm acquired a product called Kid’s Studio. I was involved in porting it to Windows and producing version 1.3.1 for both Macintosh and Windows. These versions were released on a hybrid CD-ROM, Storm’s first CD-ROM product.

Kid’s Studio is a storytelling and creation program mainly targeted at children. It is built around the concept of “the page” and lets you place objects on it. Objects can be photos from PhotoCD, drawings made with the built-in paint box, or images from a large collection, provided as Treasure Chests. Stories consist of multiple pages, and every object on the page can have a sound associated with it. Transition effects between pages are provided during playback.

EasyPhoto

EasyPhoto was Storm’s next step down the pyramid from high-end users to the consumer level. We wanted to create a product that enables consumers to manage and use photos on their computers. The consumer goal meant that we had to make a product that was even easier to use and required fewer computing resources. We also decided that this had to be a PC/Windows product because that is the machine most consumers have at home.

Although EasyPhoto was in many ways based on Photoflash, it is quite a different product. It introduces the film strip, or Gallery, as the means of storing photos, thereby insulating the user from the intricacies of the PC file system. EasyPhoto also made it very easy to use photos in documents, once you have the photos on the computer. All you have to do is pick up a photo in the Gallery, drag it to your document, and drop it. This drag-and-drop is based on OLE2. EasyPhoto is one of the first applications to implement the more advanced features of OLE2, including a custom handler. The custom handler allows storing photos in our own custom format (JPEG, of course) for much-improved storage efficiency and complete control over all image rendering. The latter allowed us to introduce a new technology called ClearPrint(TM), designed for significantly improving the quality of printed photos using today’s consumer-level ink-jet printers.

With EasyPhoto, Storm also introduced the EasyPhoto Reader, a small-footprint, low-cost, feeder-based photo scanner. It accepts up to 5x7 photos (max scan width is 4.1”) and scans them with the touch of just one button.

EasyPhoto for Macintosh

My team and I completed the Macintosh version of EasyPhoto in November of 1995, and it shipped in early December. The Macintosh version is substantially the same as the PC version, but lacks OLE support. A logical substitute at the time would have been OpenDoc, but our shipping schedule was ahead of Apple’s, so there is no OpenDoc support. EasyPhoto for Macintosh is completely scriptable and recordable, though. Check it out!

At the same time, we introduced a Macintosh-compatible version of the EasyPhoto Reader. It uses the Macintosh serial port (any port available), but is otherwise identical to the PC version described above. Performance of this model is slightly better than that of the PC model due to our use of a high-speed synchronous communication mode of the serial port.

PhotoDrive

PhotoDrive is a small photo scanner, similar in functionality to the EasyPhoto Reader. The big difference is that while the EasyPhoto Reader is an external unit, the PhotoDrive is designed to be installed in a standard drive bay of your PC. Currently, Hewlett-Packard is shipping one of its Pavilion line home PCs (Pavilion 7130) with this drive pre-installed. The drive provides up to 400 dpi resolution (twice that of the EasyPhoto Reader) and maintains adequate performance through an ISA bus interface card.

Storm PageScan USB

Storm PageScan USB is a full-page sheetfed scanner that interfaces with the computer via the Universal Serial Bus. The technology for this scanner was acquired when Storm acquired Logitech’s scanner business. Some improvements were made, and color and branding were changed before reintroducing this device to the market. The scanner is bus-powered, leaving just one cable to connect for installation. Installation is completely automatic (plug-and-play) because the necessary drivers are included with Windows 98.

Netfreight.com a.k.a. Cyntric

Netfreight.com was founded in May of 1999 by Joe Agliozzo. I joined soon thereafter, first as a consultant, and, as of September 1, 1999, became a full-time employee with the title of CTO and Vice President of Engineering.

NetFreight.com was renamed Cyntric, Inc. in May of 2000. Cyntric was developing a B2B Internet marketplace and application for procuring and managing transportation. Its primary focus was on the TL (truck-load) and LTL (less-than-truckload) business, but ultimately, expansion into other modes of transportation was planned.

On June 9, 2000, Cyntric failed to raise additional funding and closed its doors. It is for this reason that I can not point you to a website where you can check out my work. Besides managing and overseeing the development, I was technically involved with Website design, database (Oracle) schema design and maintenance, networking (design and maintenance, including firewall), and UNIX (Solaris) system setup and maintenance.

CertifiedTime, Inc.

From July of 2000 until July 2001, I worked for CertifiedTime, Inc. as CTO and Vice President of Engineering. CertifiedTime, Inc. closed its doors in July 2001 due to a lack of funding.

CertifiedTime offered its services as a Trusted Timing Authority (TTA). A service that could be used to accurately set the time on a device, computer, or application, while retaining proof of this event (including clock offset at the time). All this was done in a secure and trusted manner, with all evidentiary information held by us as a trusted third party. While mechanisms to synchronize the computer clock over a network have been widely available (e.g., NTP), mechanisms that do so securely while retaining evidentiary information have not. Having “trusted” time available through a TTA is extremely important for non-repudiation. If and when necessary, CertifiedTime can present data in court proving that a computer or device was on time or within certain margins at any given time during the use of the service. This, in turn, would be used to defend against disputes arising around the time of an electronic transaction.

At CertifiedTime, in addition to building the technical team, I was responsible for quickly delivering both a fully redundant, 24/7 version of the timing centers and a lightweight version for early deployment in new, but low-volume, geographic business areas. I deployed timing centers in Exodus centers in New Jersey, California, and Tokyo. Another deployment took place in facilities owned by Deloitte and Touche, for demonstration purposes. Both Tokyo and California deployments were fully redundant systems, utilizing extensive setups with Cisco and SUN equipment. The database used with Oracle8i.

Because the team was small, I was intimately involved at the technical level with the installation, management, and security hardening of all networking equipment, the SUN servers, the Apache Web server, the mail system, and the Oracle8i database.

I was also heavily involved in major business deals with potential partners in Europe (a major telecom firm) and with our partner Amano in Japan, where I further enhanced my prior experience doing business with Japanese firms.

Starfield Consulting

After leaving Storm, and again after leaving Cyntric and CertifiedTime, I worked as a one-person consulting firm, Starfield Consulting.

Former and current clients include:

Patents

**1**: US Patent [1995000554384](http://www.patents.ibm.com/details?pn=US05682441__) - Method and format for storing and selectively retrieving image data. Granted 10/28/1997.
**2**: US Provisional Patent Application - Methods and Apparatuses for Organizing Shipments. Filed April 2000.
**3**: US Provisional Patent Application - Trusted Timing Authority Implementation. Filed January 2001.

List of publications (incomplete)

**1**: A. G. Starreveld, W. G. P. Mooij, W. E. van Waning, and L. O. Hertzberger, "Design and evaluation of a parallel implementation of Prolog," in Parallel processing and Applications (E. Chiricozzi and A. d'Amico, eds.), (L'Aquila, Italy), pp. 457—480, Elsevier Science Publishers, Sep 1987.
**2**: T. Papathanassiadis, A.G. Starreveld, H. Bheda, A. Ligtenberg, "A Real Time Single Board Implementation for the Baseline Color Still Picture Compression Standard," proceedings of the PCS 90 (Picture Coding Symposium 1990), Cambridge, Massachusetts, U.S.A., March 1990.
**3**: A. van Inge, L.O. Hertzberger, A.G. Starreveld, F.C.A. Groen, "Algorithms on a SIMD processor array," In: NATO Advanced research workshop on multisensor fusion for computer vision, Vol. F99, J.K. Aggarwall (ed.(s)), Springer-Verlag, Berlin, 1993, pp 307-322.
**4**: P. H. Hartel and A. G. Starreveld, "Modula-2 implementation overview," J. Pascal, Ada and Modula-2, vol. 4, pp. 9—23, Aug 1985.