Hermes Wiki

MLOps

Operationalizing ML: model versioning, deployment, monitoring for drift, retraining pipelines.

Why we need this / what value this brings

A model that isn't monitored will silently degrade (drift) without anyone noticing until outcomes get visibly worse.

When to use this

Once a model is in production and its predictions actually drive decisions.

How to use or implement this

Version models alongside the code, monitor prediction distribution over time, and have a defined retraining trigger.


Empty folder — drop notes, links, and findings here as you research.

Hermes Wiki