Two different feature scale modeling frameworks are utilized for the study of aluminum (Al) deposition profiles inside micro-trenches. The first framework, which is applied in metal-organic chemical vapor deposition (MOCVD) of Al, couples a ballistic model for the local flux calculation, a surface chemistry model, and a profile evolution algorithm. The calculated conformity of the deposited film is compared with experimental results corresponding to Al MOCVD from dimethylethylamine alane (DMEAA). The outcome of the comparison is that the effective sticking coefficient of DMEAA is in the range of 0.1 - 1. There is also a strong indication that surface reaction kinetics follows Langmuir - Hinshelwood or Eley - Rideal mechanism. The second framework, which is applied in physical vapor deposition of Al, implements 2D molecular dynamics (MD) simulations. The simulations are performed in a "miniaturized" domain of some hundreds of Angstroms and are used to explore micro-trench filling during magnetron sputtering deposition of Al on a rotated substrate. Most of the experimental results are qualitatively reproduced by the MD simulations; the rotation, aspect ratio, and kinetic energy effects are correctly described despite the completely different length scales of simulation and experiment. The sticking probability of Al is calculated 0.6 for the conditions of the experiments.