[sdiy] Help for a budding young engineer...

[Olivier Gillet]

The hot topics of today might no longer be hot topics in 5 years, so
"future proofing" for 5 years might lead to a totally different
strategy from a long term "future proofing".

Things that are going to change: languages, platforms, technologies.
Things that are not going to change: mathematics, laws of physics,
modeling, methodology (how to troubleshoot? how to discover things
from experiments? how to digest information and teach yourself?)

I personally think that learning Matlab, C++, or SPICE as part of a
classroom setting is a waste of time - it makes more sense to me to
learn this through an internship, a research project, or, above all, a
personal project. I am several times more likely to hire people who
show on their CVs that they worked on a personal project during their
studies, or took part in university challenges (robot cup, solar car
racing...). The more "theoretical" classes are a long term investment
(what if in 20 years quantum computing becomes widespread?) but also a
way of increasing your learning capacities - making you more agile at
learning stuff by yourself, at spotting analogies between fields, etc.

So don't bother too much about the "relevance" of university classes
in the present technological landscape. Go for the challenging things.
But keep tinkering with present-day technology for personal projects /
internships / extracurricular university projects.

Oh and don't hesitate to totally change your plans just to have the
chance of working along a great student, or attending the classes of a
great instructor.

The 2 best classes I had while at university are automata/language
theory - just for the beauty of it and because the instructor was
truly inspiring - and an inference/learning theory class - the art of
looking at input/output pairs and figuring out what kind of processes
could go inbetween is relevant to almost everything I do.

Olivier