A Meta-Computing Architecture for Adaptive Optimization of AI Workloads in Distributed Systems

This paper proposes a Meta-Computing Architecture (MCA) for adaptive optimization of AI workloads in distributed systems. The framework introduces a layered architecture integrating monitoring, analysis, and control mechanisms for runtime optimization. An Adaptive Meta-Optimization Algorithm (AMOA) is proposed to dynamically adjust computational parameters based on workload behavior. Two evaluation metrics, Workload Complexity Index (WCI) and Meta-Efficiency Score (MES), are introduced to quantify workload complexity and performance improvement. A mathematical model and simulation-based evaluation demonstrate that the proposed framework improves execution efficiency and resource utilization compared to static computing approaches. This work is presented as a conceptual preprint and aims to establish a foundation for future research in self-optimizing distributed AI systems.