Laser based Powder Bed Fusion (PBF-LB) is often used to build complex, topology-optimized geometries to make maximum use of the freedom in design that PBF-LB offers. However, the often complex component geometry can significantly impact local thermal history of PBF-LB material during the PBF-LB process, resulting in varying microstructure and mechanical properties within PBF-LB components. Therefore, mechanical properties can locally deviate strongly from the used design values, which can be problematic for critical aerospace applications [1]. To fully leverage the freedom in design and optimization offered by the PBF-LB process, it is essential to predict and mitigate these geometry-induced variations in the microstructure and mechanical properties without restricting the design possibilities. In previous work, this problem has been addressed by introducing and optimizing variable interlayer time (ILT) based on low-fidelity transient thermal finite element models (FEM) [2]. This study expands the variable scan parameter optimization framework by combining thermal modelling approaches on multiple scales, optimizing local laser power parameters and ILT. In this optimization framework, FEM is used to determine interlayer temperatures and optimize ILT and local laser power, while an analytical model is used to predict and avoid overheating at a scan track level. The models have been calibrated and experimentally validated with the use of a longwave infrared (LWIR) camera, as well as the analysis of melt-pool geometries. A combination of LWIR and near-infrared (NIR) camera systems have been employed to monitor the stability of the PBF-LB process with and without optimized variable parameters to validate the effectiveness of the variable parameter optimization framework. The numerical and experimental results show that the variable parameter optimization framework can successfully avoid heat accumulation at component and scan track scale, improving the thermal process stability and homogeneity.