Geoengineering may become an essential tool for reducing the risks of climate change, but we do not fully understand how to develop geoengineering strategies that minimize a number of important risks.  A negative radiative forcing could be introduced, for example, using stratospheric aerosols or marine cloud brightening to offset some of the radiative forcing due to anthropogenic greenhouse gases.  However, there are many reasons for concern with any such approach, ranging from concern over “winners and losers” that might result from regional inequalities (i.e., who gets to set the thermostat?), to the irreducible uncertainty regarding the climate effects.  Much of the (modeling) research to date has imposed some particular geoengineering strategy and then looked at the effects, rather than designing a strategy to achieve a particular objective.  I will discuss two aspects of geoengineering that can be better designed with engineering tools.  First, can we optimize the distribution of solar reduction in space and time to minimize the regional inequalities?  That is, if we ever implement geoengineering, we should at least do it intelligently!  Second, how could we manage geoengineering given the deep uncertainties in climate response?  Could we test geoengineering?  Can we use feedback of the observed climate state to obtain a desired climate target despite significant uncertainty?  I will show recent results illustrating that optimization and feedback can help mitigate some of the downsides of using geoengineering.