Designing silicon photonic integrated circuit elements can be incredibly challenging and non-intuitive.  During his time at UD, he and his lab group pioneered the use of inverse design techniques, including genetic algorithms and particle swarm optimization, to design basic photonic elements.   As it turned out, the combination of these techniques with fast solvers allowed the group to produce world-leading designs for a number of passive components, including Y-junctions and waveguide crossings.  In the ensuing years, the use of these techniques to optimize the performance of passive silicon photonic components has become standard within the industry.