There is great interest in generating a terawatt (TW) hard X-ray free electron laser (FEL) that will enable coherent diffraction imaging of complex molecules like proteins and probe fundamental high-field physics. A feasibility study of producing such pulses was carried out employing a configuration beginning with an SASE amplifier, followed by a "self-seeding" crystal monochromator, and finishing with a long tapered undulator. The undulator tapering profile, the phase advance in the undulator break sections, the quadrupole focusing strength, etc. are parameters to be optimized. A genetic algorithm (GA) is adopted for this multi-dimensional optimization. Concrete examples are given for LCLS/LCLS-II systems. Show Partial Abstract