Semiconductor manufacturing is one of the most complex manufacturing processes in the world, producing chips at high volumes at the nanometer scale. Recently rapid changes in the market have shown the importance of fast, highly adaptive, and optimized semiconductor production scheduling that makes these manufacturing processes resilient to systemic shocks and disruptions.
We present how minds.ai optimizes fab-level production scheduling using two main ingredients: production-grade state-of-the-art Deep Reinforcement Learning and Scalable Confidential High-Performance Computing with Azure.
Deep Reinforcement Learning quickly and dynamically generates fab-level schedules using a user-defined mix of high-level KPIs, e.g., related to throughput, utilization, cycle times, and on-time delivery. This process is designed to augment the fab operators’ current workflow in order to safely bring the schedules into production.
In addition, HPC workloads play a critical role in enabling this AI-based revolution in manufacturing. our partnership with Microsoft Azure unlocks this revolution with highly-scalable cost-effective High-Performance Compute infrastructure, while securely and privately processing data using Azure Confidential Computing
.jpg "Sumit Sanyal, BTech photo")
