Delivering parcels from the distribution hub using Unmanned Aerial Vehicles(UAV)/Drones for last-mile delivery is gaining popularity for a smart city. The total path length of the drone while delivering at multiple locations in a single despatch depends on the order of traversal of those points. Determining the optimal tour in this context is the classical TSP (Traveling Salesman Problem) problem and is NP-complete. Since the drone has a limited capacity c, we need multiple tours, each traversing at most c delivery points. Also, the length of each tour cannot exceed the maximum flight length of the drone. In this paper, we have formulated an ILP that provides optimized multi-tour paths for the drone under these constraints. However, using the ILP solver to get the solution for large instances in real time is infeasible. Hence, our approach to solving this problem is to use an ensemble of two polynomial time heuristics—Maximum Distance Minimum Deviation (MDMD) and Highest Distance Reduced Next (HDRN). Experimental results show that our algorithm gives solutions quite close to the optimal solution from the ILP.