Genetic algorithm is often used to find the global optimum in a multi-dimensional search problem. Inspired by the natural evolution process, this algorithm employs three reproduction strategies -- cloning, crossover and mutation -- combined with selection, to improve the population as the evolution progresses from generation to generation. Femtosecond laser pulse tailoring, with the use of a pulse shaper, has become an important technology which enables applications in femtochemistry, micromachining and surgery, nonlinear microscopy, and telecommunications. Since a particular pulse shape corresponds to a point in a highly-dimensional parameter space, genetic algorithm is a popular technique for optimal pulse shape control in femtosecond laser experiments. We use genetic algorithm to optimize third harmonic generation (THG), and investigate various pulse shaper options. We test our setup by running the experiment with varied initial conditions and study factors that affect convergence of the algorithm to the optimal pulse shape. Our next step is to use the same setup to control coherent anti-Stocks Raman scattering. The results show that the THG signal has been enhanced.