A couple of weeks ago, I suggested a solution on SO using genetic algorithms to solve a problem of graph layout. It is an example of a constrained optimization problem.
Also in the area of machine learning, I implemented a GA-based classification rules framework in c/c++ from scratch.
I've also used GA in a sample project for training artificial neural networks (ANN) as opposed to using the famous backpropagation algorithm.
In addition, and as part of my graduate research, I've used GA in training Hidden Markov Models as an additional approach to the EM-based Baum-Welch algorithm (in c/c++ again).
