From Concept to Factory Floor: Designing Robust AI Architectures for Industrial Applications

About the Session

Building and deploying AI in industrial environments requires more than just model accuracy—it demands reliable, scalable, and maintainable system architectures that can survive under the real-world production constraints. In this session, we’ll explore how to architect AI solutions that not only function in the lab, but operate seamlessly and resiliently on the shop floor. We will dive deep into the architectural principles and design decisions that enable scalable and fault-tolerant AI systems—from data collection and preprocessing at the edge, to integration with PLCs and MES systems, to secure, on-premise model inferencing and retraining. Participants will gain access to reference architectures for the most common production use cases, including time-series anomaly detection, visual inspection with computer vision, and adaptive process control. We’ll highlight how to ensure data quality, enable low-latency responses, and support continuous improvement through closed-loop feedback mechanisms. This talk is ideal for solution architects, ML engineers, and automation professionals looking to bridge the gap between data science and operational AI. Join us to learn how to take your AI solutions from prototype to production—reliably, scalable, and sustainably.

Speaker

Nikita Golovko

Siemens, AI Portfolio Architect