Hacker
Piyush Jha
@piyushkjha
Projects
GraphOps — Evidence-Driven Incident Investigation & Safe Recovery
Graph HacksOct 2026
GraphOps is a graph-powered AI incident investigation system that helps DevOps and SRE teams diagnose service outages, trace root-cause candidates across interconnected infrastructure, and plan safer recovery actions. Traditional incident-response agents can inspect isolated logs and metrics, but often miss multi-hop dependencies, overlook counter-evidence, or recommend actions without understanding their potential impact. GraphOps uses FalkorDB to represent service dependencies, teams, deployments, observations, incidents, and diagnostic tools as connected knowledge. Its GraphRoute selection engine uses graph context, unresolved evidence gaps, tool costs, and risk signals to select the next useful diagnostic step and explain why it was chosen. GraphOps builds an evidence ledger, evaluates competing root-cause hypotheses, calculates the potential blast radius of proposed recovery actions, and checks preconditions before requiring human approval. Recovery is simulated only; the system does not modify real infrastructure. Built for SREs, DevOps engineers, and teams operating interconnected services, GraphOps demonstrates how graph-native reasoning can make AI agents more context-aware, transparent, and safety-conscious.