This thesis investigates the practical challenges and benefits of introducing Nix as an environment layer in a large, Bazel-based industrial C++ codebase. Using an Ericsson migration as the primary case study, supplemented by smaller controlled open-source experiments, the work evaluates migration cost, reproducibility gains, and interoperabilityfailures that arise when declarative environment management meets legacy build infrastructure. Empirical results show that Nix improves reproducibility and dependency transparency in controlled settings, but that hybrid industrial environments reveal recurrent boundary problems: Bazel sandbox visibility, interactions with rules_foreign_cc and Autotools, ABI mismatches with host libraries, and wrapper scripts that reset the environment. The thesis presents a taxonomy of migration barriers, a reproducibility baseline from controlled experiments, and practical guidance for mitigation. The findings highlight that migration difficulty is driven less by repository size and more by architectural boundary entanglement, and they provide actionable recommendations for teams aiming to adopt Nix in large, long-lived software projects.