Synthetic engineering of bacteria aims at genetically modifying existing strains in order to turn them into efficient factories for target metabolites, such as biofuels or bioproducts. Thereby, computational models are used to predict the effects of modifications and by this to enable a more directed search for promising modification strategies. In this work, we first analyzed the limiting factors of the biosynthesis, by Bacillus subtilis 168, of surfactin which is considered as one of the best biosurfactants known so far. This lipopeptidic compound is synthesized by an assembly line machinery constituted of four NonRibosomal Peptide Synthetases (NRPS) encoded by the same operon. Three potential bottlenecks were identified: the intracellular level of synthetases, the transformation of apo-form of synthetases in holo-form by a phosphopantetheinyl transferase and the intracellular concentration of surfactin precursors. First, a foreign Bacillus gene encoding an efficient phosphopantetheinyl transferase (Sfp+) was introduced into the chromosome of B. subtilis 168 sfp0. Then, the surfactin operon being regulated by a complex mechanism induced by quorum-sensing, its promoter was replaced by another promoter which led to a constitutive production of surfactin. For the third limiting factor, two strategies were performed. On one side, a knock-out of genes involved in the biosynthesis of another NRPS complex was done to limit the competition for precursors between the production of this compound and the biosynthesis of surfactin. On the other side, a new computational modelling approach based on abstract interpretation and constraint solving was developed to predict genetic modification strategies for the increase of NRPS precursors biosynthesis.