We propose a novel Interactive ML framework for advanced manufacturing. We combine incremental learning (few-shot learning and active learning) and human know-how (human-in-the-loop) to face data scarcity in quality control applications, from predictive manufacturability to predictive maintenance.