The objective of Use Case 1 (UC1) is to use machine learning to orchestrate network slicing. The SFI2 architecture, derived from the NECOS project architecture, supports the Use Case deployment. Architectural components must be explicit about embedding machine learning (ML) techniques and methods in the SFI2 architecture. In this context, the proposition is to extend the NECOS architecture to explicitly deal with machine learning techniques for slice orchestration and other phases of the network slicing life-cycle.