Imantics provides a cloud-based platform specializing in semiconductor device fabrication through IoT. Catering to large-scale fabs with diverse manufacturing automation solutions, Imantics' solutions could be deployed on-premise, as SaaS, or in a hybrid cloud model. Following their initial focus on leveraging IoT for improved yields and process uptime, they looked for new ways to scale up leveraging of insights from their installed base to facilitate further gains in time to market for customers.
The Challenge
Semiconductor fabs operate on meticulously timed end-to-end manufacturing flows. Any interruption, even a brief five-minute machine malfunction, can halt the entire pipeline, leading to significant financial losses and destruction of work-in-process material. With the rise of network-connected devices, competition has intensified, placing new pressure on yields in the face of more complex devices and semiconductors with even more stringent requirements. Imantics, relying solely on IoT, faced challenges in preemptively identifying and addressing potential malfunctions in fab lines.
The Solution
A crucial advantage an AI-driven approach to improving a baseline set of data and analytics brings is accelerating the cycle time-to-insights far faster than humans ever could.
- Transition to AI-Driven Analytics: CloudGeometry transformed Imantics' cloud platform, integrating AI to analyze diverse data types and input streams. This new capability enabled predictive malfunction alerts and real-time preventive measure recommendations.
- Deep Learning Integration: AWS Sagemaker was employed to train deep learning models on historical IoT data, enabling the prediction of equipment failures. These models, along with MLOps pipelines that detected drift and enabled continuous training, refined their accuracy over time.
- Real-time Anomaly Detection: Kinesis stream processing was introduced to detect anomalies in real-time IoT device payloads, offering operators early warnings before potential malfunctions.
- Smart Data Processing: AI algorithms were implemented to decide optimal data processing methods, choosing between Spark in-memory transformations or batch processing via EMR.
As Imantics onboarded different manufacturers with diverse IoT data capture requirements, CloudGeometry's CI/CD solutions allowed more frequent, more granular updates, streamlining the onboarding process and cutting setup the time between runs for different devices or requirements.
Better integration of analytics in all the key steps of IoT enabled manufacturing process monitoring technology significantly increased the value of Imantics solutions, turning potential insights into practical improvements that are immediately applicable to semiconductor manufacturing customers
The Benefits
The introduction of AI by CloudGeometry transformed Imantics' development and deployment processes. With the capability to adapt to data stream changes and refine machine learning models continuously, Imantics' customers experienced unprecedented improvements in yields. The shift from a solely IoT-based platform to an AI-driven approach ensured predictive and real-time equipment health checks, minimizing downtime and maximizing efficiency.
<div class="case__txt--cols"><div><h4>Cleaner ML-ready data updated continuously</h4><p>Stream machine-originated data into the platform, define data transformation and processing rules.</p></div><div><h4>Closed-loop development for IIoT</h4><p>Fully automated DevOps pipeline; configure & deploy new device payloads in minutes, not weeks.</p></div><div><h4>Fast, frequent, and repeatable</h4><p>Integrated CI/CD automation for frictionless, agile business innovation.</p></div></div>
