In the last decade, first-principles approaches to thermal transport have emerged as powerful tools to obtain predictive estimates of the thermal conductivity of crystalline semiconductors. User-friendly solvers and interfaces to first-principles codes have been developed and released to the public, helping popularize those techniques. However, this is only the first step on the way to modeling thermal transport in real semiconductors of technological interest. The conductivity of the single crystal is an upper bound to the value found in actual samples, where crystallographic defects and interfaces affect phonon scattering. The components needed to analyze such multiscale problems are far less developed, with most studies still relying on parametric relations with limited predictive power. Improving the situation is a key objective of H2020 project ALMA (www.almabte.eu), which aims to develop a software framework for thermal conductivity calculations in more complex systems. This talk will be devoted to presenting the capabilities of the code and discussing the simulation techniques implemented therein. Examples of applications will include the cross-plane thermal conductivity of crystalline and solid-solution thin films, phonon scattering by several types of localized defects, and optimization of III-V superlattice profiles.