Background <p>Traditional creep testing takes thousands of hours and incurs large costs. Calibrating constitutive models for creep offers the potential to perform shorter tests and extrapolate to longer durations.</p> Objective <p>This work focuses on the design of a complex specimen geometry, suitable for a Materials Testing 2.0 (MT2) style creep test, that uses a heterogeneous stress field and inverse identification to determine constitutive model parameters.</p> Methods <p>A digital toolchain comprising of finite element simulation, image deformation, 2D Digital Image Correlation (DIC) and inverse identification has been used to assess candidate geometries. The open source material modelling code New Engineering Material model Library 2 (NEML2) has been used to create a GPU-accelerated Virtual Fields Method (VFM) for inverse identification.</p> Results <p>Promising specimen geometries have been identified from a grid search of a simple two variable geometry parameterisation. Investigations of the design space suggest that wide stress ranges and high constitutive model parameter sensitivities drive accurate creep constitutive model identification. However, the maximum achievable stress range and sensitivity is limited by the ability of the DIC system to resolve the strain field.</p> Conclusions <p>The results demonstrate the necessity of including DIC within the design loop for MT2 tests to generate realistic and measurable specimen designs.</p>