Optimal strategy seems to be work at FAANG while iterating on side projects until something sticks.