A chip to package interaction risk assessment platform has been developed using finite element analysis, meta-modeling and genetic algorithm optimization method to tackle increasing variations in package specifications. The results show that the meta-model can efficiently predict BEOL peeling stress and solder von Mises stress of FCBGA device at reflow condition with eleven design variables. Baselines for two critical stresses are determined from qualification and mass production experiences. Room for stress mitigation is also investigated.