Aerodynamic shape optimization of the LTA system requires very high computation time and cost. Traditionally, optimization involves sampling diverse data points from design space and performing multiple experiments or CFD simulations. In the existing surrogate modelling techniques, certain sampled points can be redundant and fail to provide adequate information about the design space, thus reducing the efficiency of the process. In this paper, we propose an automated active learning—CFD integrated surrogate model, coupled with Particle Swarm Optimizer (PSO) to achieve optimal shape of an airship envelope for desired design space. This active learning approach enables the model to utilize sampling techniques that consider both the design and output space to ask meaningful unlabeled queries that better understand the design domain. These unlabeled queries are then labeled using automated CFD simulations. In each iteration, the model actively trains on these strategically selected samples such that it achieves greater accuracy with fewer training instances to learn. The surrogate model developed within this framework demonstrates a mean absolute percentage error of less than 3%, showcasing superior performance compared to random and latin hypercube sampling. This paper provides the methodology and results of fully automated shape optimization of a 2D airship hull envelope, which significantly reduces the computational time and cost.