Getting Work Done in Multiples

As some of you know, I am currently at Sumitomo Bakelite on internship, which started for me on Monday. What Sumitomo Bakelite does, is develop plastics used in IC packaging. 40% to 60% of all ICs fabricated use Sumitomo's plastics to package them. Though as benign as plastic packaging sounds, packaging for IC chips is a significant engineering problem.

What happens as temperature rises, say due to an operating chip, stresses are induced on the chip by the package because of mismaching of thermal coefficients between the chip and the plastic casing. The induced stress can lead to breaking of circuit wiring rendering the chip useless. It seems that Sumitomo Bakelite has done a good job of capitalizing on this problem to make a product and provide the consulting necessary to meet the demands of the IC industry. The way they do it is through a lot of testing and simulation.

I joined the simulation and evaluation team of this company and currently working on some low level work. Mainly, data taking and the operation of equipment for 2 experiments. As I learned from the guy that ran these machines, what he basically did was flip a switch, heat up a sample and hit "enter" on a computer for 4 hours straight in 5 to 10 second intervals. I ran the experiment twice and if I were to do it as a job, I'd go batty. They had scheduled me to run these experiments a few more times... and I thought there ought to be better ways of doing it.

What I did instead was proposed to the team that I make a program to automatically hit "enter" every 5 to 10 seconds and have the computer remotely managed some somewhere conveniant... (ie my desk). After a few inital false starts, I now have the program working today. Now 4 hours of work can be done with about 40 minutes of supervision. Wouldn't it be lovely if I could goof off for the other 3 hours and 20 minutes and still get paid for it?? (Note, I'm paid peanuts for this job! But it's kind of fun).

The next trick which I completed today was writing a little software package would let me strip temperature data out of some recorded files. What these people do was they get a guy to open every file and type that temperature value down for every file. One data set can run upwards of 300 to 400 files. I got through about 60% of a data set in about 2.5 hours and they had 10 data sets for me to get the temperature data. Riiight, can you say "slave labour"??

Well, it tuns out through a stroke of luck that I figured out how and where the temperature data is stored in these files in binary form after about half a day of searching. Once I had that down, all I had to do was write a program that would go into the directory, get a list of all the files, strip the temperature data and dump them convienantly into a text file. In 2.5 hours, a working program was running and I pumped out a script to gather data from 10 directories. The data was stripped out in about 30 seconds using the program. They also told me that they asked the equipment vendor they could do what I did... the vendor told them that it was impossible (ha!).

I still have to do a few more things by hand but it seems that I've gotten all the long and boring work done. All I have to do is strip out some more data, do some calculations in Excel and I'll have the bulk of my work wrapped up nice and tidy. Apparently, I've given these people the problem of finding more work for me to do.

What you should walk away with from this story:

Being paid by the hour sucks if you can work multiple speeds faster than what people can currently do. If I do several days worth of work in a single day my time should be worth at least multiple times what the regulars are paid, because the work was done and time was saved. So here's the question, how do you ask your boss for a 300% pay raise? There is an answer, but I'll let you think about it.