This paper presents a numerical workflow designed and developed for autonomous synthesis of vertical axis wind turbine (VAWT) blades for maximum annual energy production at a specified location given by specified wind speed distribution and prescribed tip speed ratio. The workflow can synthesize shapes of both classical VAWT designs: Darrieus and Savonius rotors. This is achieved using a novel shape parameterization scheme based on B-splines which represents a compromise between shape generality and the multitude of shape variables. The developed computational framework enables the optimizer to synthesize and evaluate a variety of geometrically and even topologically different shapes such as the Darrieus and Savonius types. Moreover, the workflow can invent (i.e. numerically generate without a resembling initial shape) new generic shapes for custom operating conditions. Both single wind-speed and systems related to real-site operating conditions specified by a given distribution of wind speeds are considered. The developed workflow consists of efficient geometry parameterization, a genetic algorithm based optimizer and a computational fluid dynamics based simulator. The rather promising results of custom-shaped vertical axis wind turbines for maximum annual energy production at a given specific site are presented using a set of case studies. (C) 2017 Elsevier Ltd. All rights reserved.