Thermal simulation of the laser powder-bed fusion (LPBF) process is a multiscale problem that requires a high-resolution representation of the melt pool region, combined with accurate prediction of the thermal field in a surrounding part with complex geometry. Purely physics-based models, such as the finite element method (FEM), are typically not efficient enough to give full-part simulations at the computational times needed for effective process optimization or control. On the other hand, the large space of possible part geometries, with an effectively infinite set of possible shapes and arrangements of features affecting thermal history, makes it difficult to train data-driven models to accurately predict temperature throughout the part. In this work, we combine a fast physics-based solver at the part scale coupled to a data-driven representation of the melt pool scale to achieve very efficient whole-part process simulations in a new method called HASTE: Hybrid Algorithm using Surrogate Temperature Evaluation. We build on our previously developed framework GO-MELT (GPU-optimized multilevel execution of LPBF thermal simulations) [1], which uses a multilevel mesh representation to couple thermal FEM solutions at melt pool, meso-, and part scales, and we take advantage of the observation that the melt pool solution varies within a well-bounded space of possible solutions and thus lends itself well to data-driven modelling. We replace the melt pool FEM solution with a dynamic model decomposition with control (DMDc) formulation [2] that is trained on a moderate number of full FEM simulations. The reduction in computational time for the melt pool temperature, coupled to a flexible but coarser-mesh FEM representation of the part scale, significantly reduces the time needed for full part-scale thermal simulations. The accuracy and performance of HASTE are demonstrated on real part geometries with scales of several centimeters. REFERENCES [1] Leonor, J.P. and Wagner, G.J. GO-MELT: GPU-optimized multilevel execution of LPBF thermal simulations. Computer Methods in Applied Mechanics and Engineering. (2024). 426:116977. https://doi.org/10.1016/j.cma.2024.116977 [2] Proctor, J.L., Brunton, S.L. and Kutz, J.N. Dynamic mode decomposition with control. SIAM Journal on Applied Dynamical Systems. (2016). 15(1):142–161. https://doi.org/10.1137/15M1013857