Due to sustainability, the Transport & Mobility sector has started a transformation to electrical machines. We show how non-parametric topology and shape optimization improves the design of the electrical machines. At present, mainly parametric optimization tools exist for optimizing the electrical machines. Overall, we suggest applying non-parametric topology and/or shape optimization for improving the design of the electrical machines with respect to torque performance, NVH (Noise, vibration, and harshness) and structural performance. The present work suggests non-parametric optimization techniques to be included in the design process for electrical machines to improve the performance. The recommended procedure would be to use topology optimization in the conceptual design phase. Then afterwards apply shape optimization based upon the newly created topology. To describe all steps of the process in detail would go beyond the scope of the talk and therefore, we concentrate on topology and shape optimization. Both build on numerical finite element analysis to solve the primal formulation for the considered physical domains, i.e., electromagnetic, structural mechanics. The optimization technology itself uses mathematical programming where the key implementations are the adjoint sensitivities for all design responses, e.g. low frequency electrical magnetic properties as torque mean, torque ripple and lumped radial forces as well as structural properties as stiffness and strength. In order to reduce special harmonic components of the lumped radial tooth force a dedicated design response is implemented. To illustrate the two non-parametric optimization methods, we will show topology optimization and non-parametric shape optimization for a reluctance and a permanent synchronous machine.