Design how computer programs can work together to solve problems (currently, problems in image processing). This often comes down to “what sort of language would make this problem easy to solve?”

The task of the architect is to break a large problem up into a lot of simpler problems that area easier to solve, much like a building architect breaks up the complex problem of making a building into work for cement workers, bricklayers, plumbers, plasterers, etc.

My current work is a large set (over 600) of programs for processing images – finding bright spots, filtering, converting into different formats, working out how two different images relate to each other – which have to work together to achieve an end task – perhaps finding ships in an image taken from a sattelite. And the images are often huge – say 900 million pixels (a small camera might be 5 million, a good camera 15 million, so these are really big).  One of the problems is how to process such large pictures quickly. The computers we work with may have 128 “cores” – like 128 separate computers in the same box; in contrast, my home computer has 2 cores (it is rather old) and my new desktop computer at work has 12. We process the image in “tiles” – think like on your bathroom wall – assigning each tile to a core. But writing programs to do parallel processing (a technical name for this, sometimes called “multi-threading”) is very difficult and easy to get wrong. Our system solves this by running them as separate programs, with messages telling each one which parts of the image it should process; the operating system (we use Linux) is very good at this and knows how to get it right, all our programs must get right is the dividing up of the image. The tiles we use overlap a little bit so we do not miss anything in the joins.

It might just be sitting in front of my computer most of the day designing (and writing) software.

It usually would involve talking to colleagues, both here and overseas (by email, video conference, sometimes by traveling) about issues they are having.

Often it involves just sitting and thinking (or walking around and thinking) about the best way to solve a particular problem, or about a better way to do something.

As well as looking for better ways to do the task, I am also the person that gets handled the urgent or “too hard” problems by other members of the team.

I am involves in the CSIRO Scientists and Mathematicians in Schools program, and the schools I am involved with could use the money for equipment for their science rooms or support for their science clubs.