How would you adapt your previous research in optimization or reinforcement learning to solve real-time vehicle routing problems for Meta Logistics?