International Journal of Science, Engineering and TechnologyAuthors: Emily Carter, James Whitaker, Lauren Mitchell, Benjamin Harris, Ryan Sullivan, Ananya Kulkarni
Abstract: Enterprise Resource Planning systems have traditionally relied on batch-based integrations and tightly coupled extension models that limit responsiveness in dynamic organizational environments, particularly within human capital platforms where workforce data changes continuously and decisions are time-sensitive. This study proposes a streaming-native ERP extension architecture that leverages Kafka Streams, event-driven microservices, and scalable big data infrastructures to enable real-time, intelligent decision automation across human capital management ecosystems. Rather than treating ERP extensions as static transactional add-ons, the framework reconceptualizes them as continuous event-processing layers capable of ingesting employee lifecycle events, compensation updates, time and attendance signals, performance interactions, and compliance triggers as live data streams. Through stateful stream processing, schema-governed event pipelines, and domain-aligned microservices acting as autonomous decision agents, the architecture supports low-latency anomaly detection, automated approval routing, predictive workforce analytics, and policy-driven compliance enforcement. A distributed analytics layer maintains historical data persistence and model retraining capabilities, enabling adaptive learning and sustained optimization. Comparative architectural evaluation demonstrates significant reductions in decision latency, improved operational transparency, and enhanced governance consistency when contrasted with conventional batch-oriented ERP customization approaches. The findings indicate that streaming-native ERP extensions represent a structural evolution in enterprise system design, transforming human capital platforms from reactive reporting environments into proactive, intelligence-driven ecosystems capable of continuous insight generation and automated, context-aware decision support.
DOI: http://doi.org/10.5281/zenodo.18742388
