Ai Mlops

This covers the full production ML lifecycle: data ingestion with dlt, deployment patterns (batch, real-time API, hybrid), drift monitoring, automated retraining triggers, and incident response runbooks. It's grounded in Jan 2026 standards and links NIST frameworks and EU AI Act compliance where relevant. Use it when you're moving models to production or debugging why your deployed system is drifting. The skill includes decision trees for picking deployment strategies and copy-paste templates for monitoring setups. It assumes you've already trained something and now need to ship it, monitor it, and keep it working. The security section covers prompt injection and RAG attack surfaces, not just infrastructure hardening.