Role : AI / ML Solution Architect
Location: Waukesha, WI
locals Preferred.
Exp: 14+yrs
Day one onsite
Summary:
Own the end-to-end technical architecture of the hub-and-spoke platform, from the Azure cloud Hub to the on-premise Spokes, ensuring the system is performant, scalable, secure, and suitable for Computer System Validation (CSV).
Key Responsibilities
• Own the hub-and-spoke technical architecture: cloud Hub plus on-premises Spokes.
• Design the deterministic-first computer-vision inspection pipeline: OCR, barcode/UDI decoding, object and symbol detection, print-quality CV, and a fusion/decision engine that produces reproducible pass/fail verdicts.
• Define the model lifecycle: training, model registry and versioning, governance/approval, and signed distribution to sites.
• Architect the Azure Hub (Azure ML, AKS, ACR, storage, PostgreSQL, monitoring, networking) and the containerised on-premise Spoke (GPU-enabled, Docker).
• Ensure data sovereignty (production images never leave the site), security, and audit/traceability suitable for CSV.
• Meet performance targets (label processing under 3 seconds) and design for scale and high availability.
• Experience developing and integrating applications using REST APIs with manufacturing systems (eDHR / MES).
• Guide the engineering team, review designs, and drive model and vendor selection, including licensing and origin compliance for global deployment.
• Support CSV, qualification, testing, and validation documentation.
Required Qualifications
• 8+ years in software or solution architecture, including AI/ML systems.
• Strong computer vision and machine learning background: OCR, object detection, and vision-language models.
• MLOps: model registry, CI/CD, containers (Docker, Kubernetes).
• Cloud architecture, Azure preferred (Azure ML, AKS, ACR, storage, networking); on-premise or edge deployment with GPUs.
• Proficient in Python; strong system-integration experience (REST; SOAP/XML a plus).
• Experience designing distributed or hub-and-spoke systems.
Preferred.
• Regulated / GxP or CSV experience (medical device, FDA / EU MDR).
• Hands-on with PaddleOCR, YOLO / RF-DETR / Detectron2 or similar; edge-inference optimisation (ONNX, quantisation).
• eDHR / MES integration; manufacturing or industrial machine-vision experience.
• Self-starter with the ability to work independently and collaboratively.
Never miss an opportunity! Create an alert based on the job you applied for.