ZEISS MLOps Inference MediTech Platform on Azure
Overview
At ZEISS Meditec AG, within the AI Medical Technology R&D department, I contributed to the design and development of an MLOps inference platform on Microsoft Azure.
The platform was built around a central data layer for structured medical device data, enabling AI-driven decision support systems and intelligent medical applications.
The project’s mission was to establish a secure, compliant, and scalable cloud-native platform for AI in regulated medical environments — ensuring full alignment with ISO 13485, IEC 62304, and ZEISS enterprise strategy.
Role & Responsibilities
As Senior Cloud and MLOps Engineer, I designed, implemented, and guided the technical development of the Azure-based MLOps platform.
- Facilitated daily SAFe / Scrum ceremonies for project and delivery alignment
- Authored medical regulatory documentation compliant with ISO 13485 and IEC 62304
- Defined and implemented Azure cloud strategies for regulated environments
- Designed secure, compliant cloud-native architectures following Azure best practices
- Moderated architecture and requirements workshops with business and technical stakeholders
- Coordinated cross-functional collaboration with partners and internal experts
- Conducted architecture reviews ensuring quality, compliance, and continuous improvement
- Defined and delivered MLOps features, epics, and user stories for product backlogs
Applied Methods & Tools
- Project Methodology: V-Model combined with Agile Scrum for hybrid delivery
- Regulatory Compliance: Documentation aligned with ISO 13485 / IEC 62304 standards
- Architecture Workshops: Captured business, data, and system requirements
- MLOps Lifecycle Management: Implemented scalable and secure AI workloads
- DevOps & IaC: Automated provisioning and deployments for reproducible environments
- Continuous Improvement: Regular architecture reviews and technical refinements
Applied Technologies
- Azure Cloud Services: AKS (Azure Kubernetes Service), Event Hub, Service Bus, Synapse, Storage Gen2
- Kubernetes & Microservices: Python-based services using Helm charts and DAPR framework
- Machine Learning Models:
- U-Net segmentation and anomaly detection for medical image analysis
- Classification models for automated diagnostics in clinical workflows
- CI/CD & DevOps: Azure DevOps Boards, Pipelines, and Repositories for full automation
- Infrastructure as Code (IaC): Terraform modules and Kubernetes YAML manifests
- Security & Compliance: Role-based access, encrypted storage, and policy enforcement
Impact
- Delivered a fully compliant MLOps platform supporting AI model inference for medical devices
- Improved traceability and reproducibility in AI workflows through IaC and DevOps automation
- Enabled cross-functional collaboration between data scientists, engineers, and medical experts
- Established a secure, cloud-native foundation for scalable AI-enabled medical solutions
- Accelerated innovation in AI-powered diagnostics and decision support systems
Summary
The ZEISS MLOps Inference MediTech Platform showcases the integration of MLOps best practices, Azure cloud architecture, and medical regulatory compliance into one unified platform.
Through secure infrastructure design, automation, and cross-disciplinary collaboration, this project advanced ZEISS’s mission to bring intelligent, data-driven healthcare innovations to market safely and efficiently.