Fixed-wing aerial robotic technology has advanced to the point where platforms y persistent sampling missions far from remote operators. Likewise, complex atmospheric phenomena can be simulated in near real-time with increasing levels of fidelity. Further- more, cloud computing technology enables distributed computation on large, dynamic data sets. Combining autonomous airborne sensors with environmental models dispersed over multiple communication and computation channels enables the collection of information essential for examining the fundamental behavior of atmospheric phenomena. This paper describes progress toward the development of an autonomous airborne scientist for studying severe local storms. This autonomous scientist combines unmanned aircraft systems, meshed networked communication, cloud computing infrastructure, online and offine numerical eather models, and new onboard sensors. The existing system architecture will be described along with results from recent field deployments validating and assessing various subsystems.