Evaluation & guardrails
Designing evals, SLOs, and guardrails for non-deterministic agents — measuring quality where there is no single correct answer.
Engineering Manager · LinkedIn
AmarChaudhari
Making AI agents reliable enough to trust in production.
I've spent a decade making large-scale systems dependable — reliability platforms, disaster recovery, and chaos engineering across hundreds of critical services. I'm now focused on the next frontier: reliability for Agentic AI.
Where I'm headed
Agents are powerful but unpredictable. My work is making them trustworthy — applying hard-won reliability practice to a new class of systems.
01
02
Failure-mode engineering
Bringing chaos engineering, disaster recovery, and incident practice to agentic workflows so failures are expected, contained, and recoverable.
03
Agents in production
The observability, capacity, and control-plane work that lets agents operate safely at scale — not just in a demo.