| This paper considers an operator that deploys Unmanned Aerial Vehicles (UAVs) to service requests from users/devices in an Internet of Things (IoT) network; each request requires one or more so called {\em services} such as a database or an Artificial Intelligence (AI) model. A key problem faced by the operator is to determine which services to place on UAVs and which requests to transport to a base location for execution; note, UAVs have resource constraints and hence, they are unable to carry all services or/and serve all users in one trajectory. Further, they do not have connectivity to a base station during flight. To this end, given a set of requests, this paper presents novel solutions that minimize the total completion time of these requests. First, it outlines a novel Mixed Integer Linear Program (MILP) that can be used to calculate the optimal assignment of services to trajectories. MILP also decides which requests are transported to a base location for computation, and schedules the return of result(s) to users. The second solution is a heuristic method named MinTime that chooses UAVs and users according to traveling and task execution time. The results show MinTime achieved 91.15\% of the optimal result in terms of requests completion time. |