Abstract: High performance scientific applications provide very interesting challenges from the software engineering point of view. In addition to their high performance requirement, their codes exhibit a long life cycle that includes reuse as part of code-coupling applications. Dedicated programming models are required to ease the adaptation of these codes over time without introducing overhead at runtime. HLCM is a component assembly model that supports high level concepts easing code adaptation. It prevents runtime overheads by implementing these concepts through transformation applied at deployment to generate a concrete assembly. This paper deals with our experience with metamodeling and model transformation as used to implement HLCM. It provides some feedback on the advantages and drawbacks we found by using a model driven approach.