The NWO Open Competition M-2 project is called GAT4ML: Green Architectural Tactics for ML-Enabled Systems
It includes one PhD position and one 3-year postdoc position. While Green AI research, i.e., how to make AI-based systems more environmentally sustainable, has made great progress in recent years, it is still difficult for practitioners to put research results into practice, especially for more recent system types like LLM-based systems, RAG-based systems, or LLM agents. We try to address this via the concept of architectural tactics, which are reusable design decisions to improve a specific quality concern. In the project, we will a) curate an extensive catalog of tactics that respect recent advances in Green AI and newer forms of ML-enabled systems, b) improve the representation of the tactic collection to make it easy to browse and select tactics according to specific requirements, and c) gather additional empirical evidence for the tactics regarding their quality impact, prerequisites for their acceptance in industry, and effective actionable guidance for their integration in real-world scenarios. To read more, click here.