The purpose of this work is to investigate how phenomenologically successful constructions in the field of high energy physics can be embedded into a fundamental theory, and what we can learn from this procedure. In fact, the synergy of an effective and a fundamental approach might be very fruitful. On the one hand, it poses several theoretical challenges to potential candidates for a fundamental theory, which might be ruled out or restricted in their generality. On the other hand, this mutual interplay can inspire new experiments and observations or suggest correlations which are invisible from an effective point of view. Our starting point is that string theory is entitled to be a valid candidate for a fundamental theory by a series of properties such as the fact that it is a consistent, UV‐finite, anomaly‐free quantum theory and it describes, in appropriate limits, quantum field theory and general relativity, hence providing a unified description of the four known forces. Within string theory, we choose to work in the framework of type IIB flux compactifications where the issue of moduli stabilization can be successfully addressed. On the phenomenological side, we focus on particle physics and cosmology, and in particular on the MSSM and the mechanism of inflation, respectively. We construct and study two different models of inflation in string theory. These are models of brane inflation, i.e. models where the primordial exponential expansion of the universe is driven by a scalar field representing the position of a D3‐brane (a 3+1‐dimensional dynamical object) in a compact space. We show explicitly that, allowing for fine tuning, a prolonged stage of slow‐roll inflation can be achieved during which perturbations are generated in good agreement with the CMB data. In addition, we give an example of the fruitful synergy we mentioned above. One of our two models, namely inflation at the tip, might induce also DBI inflation which produces peculiar features in the spectrum of density perturbations that might be looked for using CMB data. On the particle physics side, we consider a model, the large volume scenario (LVS), which, among other things, provides a rationale for a low scale susy‐breaking. We perform some checks on the consistency of LVS as a string theory model. To this end, we formulate an educated guess for the form of string loop corrections. We then show that physical predictions such as the soft susy‐breaking terms are actually surprisingly robust against the inclusion of these corrections.