Little slow replying to this one, but that's fortunately given 3 other people in a similar position to reply and I can't agree more with the sentiment. Have I used my knowledge of approximating PDEs using multi-variate finite element methods? No, not really - though it makes for some interesting conversations with meteorologists occasionally. Whether it's the deep understanding of logical constructs (everyone should know what the contrapositive is) or the ability to model a problem and remove cruft, the way of thinking a maths degree instills will be invaluable forever.