The project introduces a localized AI framework designed to optimize semiconductor and micro-nano-manufacturing environments. The research focuses on aggregating disparate data streams from complex nanofabrication equipment into streamlined, standardized data pipelines. These pipelines feed custom-tailored compact language models that operate locally to maintain strict data security and minimize latency. By synthesizing equipment logs, sensor metrics, and operational manuals, the platform equips cleanroom technicians with intelligent, context-aware troubleshooting assistance and real-time process control capabilities, ultimately accelerating throughput and reducing manufacturing defects.
Application/demonstration
The planned demonstration will feature a live cleanroom deployment where a localized compact language model successfully processes real-time sensor streams from a nanofabrication tool to identify an active process deviation and deliver step-by-step corrective instructions to the operator.