Resilient AI Architecture for Enterprise Applications using Cloud Disaster Recovery and Fault-Tolerant Computing

Abstract

The increasing adoption of artificial intelligence (AI) in enterprise applications has created new opportunities for automation, predictive analytics, intelligent decision-making, and personalized digital services. However, the growing dependence on AI-enabled applications also introduces significant risks associated with service interruptions, infrastructure failures, cyberattacks, data corruption, model failures, and cloud outages. Conventional disaster-recovery approaches may be insufficient for AI workloads because AI applications depend on complex combinations of models, data pipelines, computing resources, application services, APIs, and specialized hardware. This study examines a resilient AI architecture that integrates cloud disaster recovery and fault-tolerant computing to improve the availability, reliability, and continuity of enterprise AI applications. The proposed approach combines redundancy, automated failover, multi-zone and multi-region deployment, continuous data replication, model version management, container orchestration, health monitoring, and intelligent recovery mechanisms. The research adopts a design-oriented methodology involving literature analysis, requirements identification, architectural design, simulation, experimental evaluation, and expert validation. Key performance indicators include recovery time objective, recovery point objective, service availability, failover latency, data consistency, model integrity, and application performance after recovery. The study argues that resilience must be designed across the entire AI lifecycle rather than being limited to infrastructure backup. Integrating fault-tolerant computing with cloud disaster recovery can provide enterprises with a scalable and adaptive foundation capable of maintaining critical AI services during infrastructure failures, cyber incidents, network disruptions, and unexpected operational events.