Finding the shortest path for an autonomous robot in static environments has been studied for many years and many algorithms exist to solve that problem. While path finding in the static setting is very useful, it is very limiting in real world scenarios due to collisions with dynamic elements in an environment. As a result, many static path planning algorithms have been extended to cover dynamic settings, in which there are more than one moving objects in the environment. In this research, we propose a new implementation of multi agent path finding setting through A* that emphasizes on the path finding through a centralized meta-planner that operates on the base of Bag of Tasks (BoT), running on the distributed computing platforms on the cloud or fog infrastructures and avoiding dynamic obstacles during the planning. We also propose a model to offer a “Multi-Agent A* path planning as-a-Service” to abstract the details of the algorithm to make it more accessible.