Calibrate-Then-Delegate: Safety Monitoring with Risk and Budget Guarantees via Model Cascades

Develop a model-cascade toolkit that allows developers to set cost/safety budgets for LLM monitoring. It optimizes when to escalate to an expensive safety expert versus using a fast local probe.

Suggested repo: gatekeeper-cascade

"Budget-aware LLM safety monitoring with cascading models."

Estimated effort: 45h