Additive Manufacturing (AM) has emerged as a competing and complementary manufacturing method, with the main advantage of enabling complex or otherwise imposable geometries to be manufactured. Although the AM processes have advanced rapidly, the development of materials tailored for AM are lagging behind. This hampers Aluminium producers from reaching new market and application with AM, such as aerospace and automotive both requiring high strength alloys. From a metallurgical perspective AM holds an untapped potential in reaching strength levels beyond traditional manufacturing methods. The potential is linked to a very high solidification rate, typical in the order of 105-106°C/s. This enables alloying elements to reach extended level of super saturated solid solutions (SSSS) far from the thermodynamic equilibrium. Thus, many elements with limited or even negligible solvability could, with AM, reach useful levels of SSSS for precipitation and solid solution hardening. To explore this new landscape of potential alloys, we will employ atomistic simulations to investigate the thermodynamic tendency of Al-based intermetallics to form precipitates as a function of temperature. We will compare the behavior of alloys that are known to undergo precipitation hardening (e.g., Al-Mg-Si and Al-Cu) with that of alloys that are known to not undergo precipitation hardening (e.g., Al-Fe or Al-Si) to gain atomic-level understanding of precipitation hardening mechanisms. The target is then to search for the signature of precipitation hardening among novel alloys. The theoretical predictions will be experimentally verified. To gather the necessary data for this comparison we will use hybrid MD / kinetic-MC / thermodynamic-MC simulations (LAMMPS). The MD / kinetic-MC approach at the basis of our investigations is described in detail in [Tavenner et al, Comp Mater Sci 218 (2023) 111929]. The kMC algorithm proposed by Tavenner et al does not need a priori-knowledge of migration rates: swaps between neighbors are accepted according to min[1, exp[-∆E/(kBT)]. However, the algorithm neglects the eventuality of vacancy migration (atomic exchange with an empty site). For this reason, we will implement a hybrid MD / kMC / tMC method. Every "X" MD and kMC steps, our algorithm will call for tMC attempts to displace atoms to random neighbor positions, which may include a vacancy site. Our MD/MC simulations will be initially based on semi-empirical interatomic potentials (eg MEAM and EAM). There are several reliable potentials available in the literature for Al-based intermetallics (https://openkim.org/browse/models/by-species). However, we will progressively shift toward development of machine-learning-interatomic-potentials (MLIP). This is especially the case for complex multinary and unexplored systems, for which no (M)EAM parameterizations are available. MLIP development will be based on VASP on-the-fly learning during ab initio MD simulations at finite temperatures. We will build AIMD supercells of intermetallic solid solutions that mimic both clustering and ideal configurationally-disordered cases. This will maximize the diversity of local chemical environments and thus optimize robustness of MLIPs to be trained on ab initio data.