Differentiation is a process fundamental to multicellularity. In its simplest form, differentiation converts self-renewing stem cells into non-proliferative cells with specified function. This process is inherently susceptible to mutant takeover - mutant stem cells that never differentiate produce excess proliferative daughter cells, driving cancer-like expansion and decreasing the availability of differentiated cells to the organism. It has been proposed that coupling differentiation to an essential trait can select against these mutants by producing a biphasic fitness curve. This would provide mutant stem cells that do not differentiate with a selective disadvantage. However, this theory has yet to be tested experimentally. Here we use "fitness landscape engineering" to design and construct a synthetic biological model of stem cell differentiation in Escherichia coli with biphasic fitness. We find that this circuit is robust to mutations as predicted. Surprisingly, its optimal differentiation rate is robust to a wide range of environmental pressures. This environmental robustness is driven by transit-amplifying cells that differentiate and proliferate irrespective of environment. These results provide new interpretations for natural differentiation mechanisms and suggest strategies for engineering robust, complex multicellular consortia.