Here are some different things I've built over the last 20 years or so.
The early years
In the summer of my 16th year, I built a halfpipe in my parents' backyard. It was 16 feet wide and 25 feet long. Strangely, I didn't make any drawings before I built it-- I just thought about it a lot and calculated the right board lengths as I was building. Building skateboard ramps was how I learned that a two-by-four is stiffer in bending on edge than laid flat.
For a few years in the late 90's, I was a high school math teacher in San Francisco. I owned no furniture, but I did have a circular saw, a jigsaw, and a cordless drill. Within a few weeks, I had built all the furniture in my apartment except a desk chair.
Some of it was weird, like this hybrid bookcase/desk.
(The skull candle sitting on top was the prize for the Best Halloween Costume, which I won at work that year.)
The desk was a clone of one I built in college.
Along with Oliver, Dorian, and some of the other solar car zealots around San Francisco, I designed and built a solar-powered car. We raced it from Washington, DC, to Orlando, Florida, in the summer of 1999. This project is what convinced me that I needed to switch professions from teaching to engineering.
A device for programming mice
This is the first device I made as an engineer for money, circa 2000. I was the inaugural summer intern at MindTribe Engineering; I spent a good eight weeks making five instances of this box, which programmed wireless mice with the codes that made them broadcast at different frequencies in different countries. It was based around the legendary Motorola 68HC11. I hand-soldered the 74HC244 output buffers needed to drive the LEDs.
Around 2002, Iowa Thin Film Technologies (now PowerFilm Solar) figured out how to make an LCD that turned clear, rather than dark, when power was lost. I built the analog drive circuitry for a motorcycle helmet that used a photocell to darken the visor as the light brightened. So far as I can tell, the product never made it to market.
A rackmount power strip controlled through the internet
The picture below shows me assembling a prototype of a massive 40 amp power strip that you can control through the internet. It's the first commercial product I worked on, and it's still for sale today, though I suspect that the manufacturer is no longer using the Rabbit 2000 microprocessor we used. This was also the first product I took through UL 60950 and FCC certification.
Laser eye surgery machine
I was the mechanical designer for the concept development stage of this product for Alcon when I was at Ideo in 2005. From what I can tell, it never made it to market; it seems that Alcon issued a massive product recall of this system's predecessor and then decided they were better off buying the next closest competitor (Wavelight) than building the machine themselves.
More recent work
For the last few years, I've been designing machines for the renewable energy industry. Most of what I've built is secret, but here are a few photos that give you a taste without revealing too much.
A cellulosic ethanol continuous fermentation system
Never before have so many months of my work hung in a position so precarious.
A test system for a large PCB, using another large PCB
Solar cell loom loader
One of the machines I'm most proud of was just shipped to China a few weeks ago (spring 2010). It loads wires into looms for making photovoltaic modules.
On the side
I bought a 1973 Porsche 914 from a guy down the street and converted it to electric power.
I wrote an open source Python library for calculating the location of the sun relative to the earth.
It's called Pysolar.
A plywood castle
After an unfortunate disagreement with the city of Cambridge about what constitutes a temporary structure for protecting your car from the weather, I ended up with some excess plywood. Hence, we built a castle.
A level floor
Our house has one room with a horribly sloped floor that needed a quick remedy. Much to the chagrin of my pragmatic friends, I measured the floor and wrote a Python script to calculate the contours of a least squares fit of a paraboloid to the measurements. I then cut layers of masonite to make a level platform that fit the doubly-curved surface.
A zipline at my brother's house
My brother discovered a grown-over grass tennis court in the woods near his house. We decided it would be best to arrive by zipline.
Sandhill Challenge car
In those heady days of the dotcom boom, when doing stupid things was smart, detritus from the MIT, Stanford, and Yale solar car teams coalesced to build a soapbox derby car. A few months before the race, we measured the slope of Sand Hill Road at night to determine the optimal weight for the car. (Strangely, this attracted the attention of the cops. How much less threatening can you get than measuring the slope of a piece of asphalt?)
With a strong pushing team at the start, you want a very heavy car (~350 kg) to win. We built ours around a massive piece of steel I-beam with a rack for lead plates underneath. Unfortunately, we overestimated the strength of our pushing team. During each of our runs, we were removing lead plates and improving our time, but it was too little, too late. I think we placed fourth.
In the picture below, you can see we are delivering a sound thrashing to a block of cheese.
The most massive piece of wood I have ever worked was this piece of ash.