MLOps Infrastructure RFP

Suppliers to develop scalable capabilities for managing the risk of ML, AI, and Agentic AI models.
- Key risks include:
• Model risk: inaccurate output such as hallucinations, inaccuracies, and bias, with potential patient harm, revenue loss, and adverse reputation impact.
• Scalability risk: inability to launch and monitor new models cost effectively and on-time.
• Explainability risk: lack of adequate capabilities for observing model inputs and output can hinder model application, trust, and accountability.
• Auditability risk: Incomplete records of model versions, data lineage, approvals, deployments, and configuration changes may hinder oversight, accountability, and investigation.
• Data protection and security risk: AI models are consumed as endpoints and have access to a wide range of data that typically includes sensitive data
• Operating risk: AI/ML models have become integrated into operational decision-making and automation processes. Failure of AI/ML models, lack of consistent availability, and inaccurate output could potentially result in patient harm, financial loss, and other adverse outcomes.