AutoInfraOps: An Agentic DevOps Coordinator for Autonomous Multi-Cloud Monitoring, Optimization, Governance, and Self-Healing Infrastructure Operations

Abstract

Multi-cloud adoption enables resilience, flexibility and vendor independence, but it also increases operational complexity through heterogeneous interfaces, fragmented moni-toring, inconsistent governance, and difficult incident response. Traditional DevOps and AIOps solutions provide monitoring, automation or optimisation in isolation, but they rarely deliver end-to-end autonomous, governed and explainable multi-cloud infrastructure operations. This paper proposes AutoInfraOps, an agentic DevOps coordinator for autonomous multi-cloud monitoring, optimisation, governance and self-healing infrastructure operations. The framework integrates specialised agents for monitoring, diagnosis, planning, optimisation, governance and remediation. It uses a multi-cloud adapter layer for AWS, Azure and GCP, an event-driven runtime for agent coordina-tion, anomaly detection for SLA violation identification, cost-performance scoring for workload placement, Human-in-the-Loop approval for high-risk actions, and SHA-256 hash-chain audit logging for decision traceability. The system is evaluated using simulated scenarios including AWS latency spike, Azure cost anomaly, GCP availability drop, cross-cloud dependency failure, cascading failure, high-risk failover and rollback events. Experimental results indicate reduced incident response time, improved SLA compliance, cost-aware remediation, and stronger governance visibility. AutoInfraOps demonstrates a practical foundation for autonomous, explainable and policy-aware multi-cloud operations.