At Azentio, I build LangGraph agents for performance, SRE and on-call teams. The Observability Agent queries Oracle, Prometheus and Grafana, then uses runbook RAG to rank possible causes; it cut time to a first root-cause lead from hours to about 10–15 minutes.
I also built guardrails and a replay harness to make agent findings verifiable and test changes without live systems. For PerfOps, I developed a LangGraph agent that runs and analyses JMeter tests unattended, with adaptive load ramp-up and SQLite checkpointing for runs that survive server restarts.
