OK, you asked for it...

I'm CTO and one of three co-founders of LandSonar - now SpeedGauge. LandSonar was founded in 2000 with the idea of taking GPS "breadcrumb" observations from vehicles that are managed by logistics companies and baking them into a predictive model of traffic patterns. We basically invented the market for predictive traffic (as opposed to "real-time" traffic reporting for infotainment). Our data was licensed by NAVTEQ to form the basis of their Traffic Patterns product, as well as several AVLs.
The traffic business is cutthroat, and we never raised enough money to be competitive. So we sold the traffic portion off to concentrate on our SpeedGauge product, which uses the same vehicle observations to do driver rating (based primarily on speeding incidents). With that model, we are growing, and profitable!

Before that, in roughly reverse chronological order:

I was the Network Services Architect in the Advanced Technology Labs at Adobe Systems. That was a pretty nebulously defined position - I was hired to help the Mobile and Devices Business Unit find their way forward by building new network services that matched their handset technology. It was, mostly, a bust - but I did build several interesting platforms for location-based applications. Pity they couldn't figure out how to make use of them. I was laid off in the great re-org of December '08, along with many of the other folks working on mobile. A good decision for Adobe, but not for the people.

I was an Area Director in the Architecture Area for the Open Grid Forum (OGF), which means that I spent an incredible amount of time on conference calls, trying to herd cats, err, volunteers. That position also got me a seat on the Grid Forum Steering Group, which gathered regularly to push documents through the standards sausage-making machine, and tried to figure out how to get more cats to herd. I was also on the OGF Technical Strategy Committee, trying to balance the conversation of what technical work would make the OGF relevant to the industry vs. what technical work was likely to get done with volunteers. It was a very interesting dance. OGF is filled with very bright and energetic people - I enjoyed my interactions with them, but I learned that the standards game is not one I enjoy playing.
Before that, I was the outgoing chair of the Enterprise Grid Alliance (EGA) Technical Steering Committee.
That was part of working at Oracle, in the Grid Technologies Group, trying to extend the notion and operation of a database to The Grid (whatever The Grid might actually be). We were working on a model for data access and propogation across organizational firewalls, as well as a load-balancing model (think about load balancing for stateless connections to web servers, and substitute stateful connections to databases). We built a mechanism for moving live OCI connections from one instance to another (across servers) and a set of heuristics to do live load balancing.

Before that, I was the CTO of Unimobile, Inc. I'm sad to say that the company went under on my watch, but I really didn't have much to say in the matter. Unimobile specialized in global wireless data solutions for enterprises - which meant tools to send timely alerts to cell phones and other wireless devices, worldwide. (Think margin calls, credit limit or bill reminders, airline flight delays - with the ability to take some action from your phone - other than just calling!) SMS messages aren't very sexy but have surprisingly high utility. Working with disparate wireless carriers was an education. Getting decent performance, a decent price point, and interesting functionality from the phone network made for fairly interesting work. It's one of the many facets of ubicomp (see below) that has slowly come into being... I also designed and deployed a system that allowed enterprises to do two-way data retrieval via SMS from "regular" cellphones - think of very simple web-based systems that don't require graphics or a computer, but allow you to get at things like your calendar, contacts, stock quotes, etc. Not nearly as "web-like" as the browser-based systems that carriers and various companies are touting, but it runs on the devices people have in their pockets. Take a look at this document for more discussion of the disconnect between phones and the web, and some possible middle ground to explore. (Unimobile was eventually bought by Verisign, and they continue to sell and operate the products and network we built.)

Before Unimobile, I designed the delivery scheduling and execution system for Webvan. There's a long story to be told there, other than what the public knows ... maybe someday someone will tell it. Probably not. I did some of my finest technical work there - unfortunately, the economics of the company didn't support it!

I spent about four years at Silicon Graphics building systems to move visual bits around very fast. First, it was animation frames on ATM networks for DreamWorks; then it was HDTV and uncompressed film across striped HIPPI ("the world's biggest and fastest Quicktime files") for anyone who would pay. Unfortunately, nobody did.

In a former life (or so it seems), I worked in the Computer Science Lab (CSL) at Xerox PARC. I spent my time there working on various aspects of Ubiquitous Computing - an idea whose time has apparently finally come, albeit not from PARC.

Before that, I worked at Digital Equipment Corporation - first as an intern at DECWRL (RIP). After I got my Ph.D. I started full-time at DECWRL and then DECWSL. I did some foundational work that supported the X Window System, worked on a personal workstation/supercomputer ("Titan"), and helped design a multi-processor version of the Titan. With Brian Reid, I built the (then) world's fastest TCP-linked PostScript printer (LPS40), but more importantly taught the VMS side of the house what the Unix market was all about (not to mention a thing or two about modular and portable code). At WSL, I helped ship the Display PostScript extension for X. You've never heard of any of it, I'd bet.

Say, how did this turn into a resume?

In grad school, I was supported by CSNet. I brought up the Purdue IMP, oversaw the transition from NCP to TCP (we were running both), built the first in-host gateway (both in Rob Gurwitz's TCP for 4.1BSD as well as in 4.1c), ran a whole lot of twisted pair wire for our proNET ring, got really really good at installing Ethernet transceivers (back in the day when you had to drill a hole in the cable), learned a thing or two about subnets and IP fragments, and helped build the IP-over-X.25 software. My thesis involved designing, building and measuring a distributed file system. It wasn't terribly interesting, but it was novel enough to get my degree, and that was all that mattered.

The networking thing began, I think, in 1979 during a talk by Brian Kernighan at the Toronto Usenix Conference - he was discussing uucp. At the time, I was at a small grad school running a pdp-11/60 and wishing I had some like minded folks to talk to. uucp was the answer. I back-ported it to V6 and started hand-dialing the Racal-Vadic 1200 modem to connect to the fledgeling network. Yow.

That was, I guess, my third Usenix - the first one was at Harvard, a week after a professor handed me the Unix Programmer's Manual and said I should read it. I think I spent almost the entire conference wondering what the hell this "inode" thing everyone was talking about was - none of my CS books had ever mentioned it!

Before that? A random walk through the IBM computers of the era, starting with an IBM 1130 in high school (I found the computer, found a book on FORTRAN in the downtown library, and wheedled my way into the computer room), a few different 360s, a stint as operator and tape monkey on a 3083 (playing with APL on the side), and a brief fling with a 1620 (the CADET - Can't Add, Doesn't Even Try). I punched a lot of cards and read a lot of listings. It taught me how to plan development and maximize test results out of minimal runs, but I really prefer interactive debuggers even if they promote sloppy work habits.

What a long, strange trip, indeed.

Last updated Dec 14 2007 by cak