Meta-Computing Architecture for Adaptive Optimization of AI Workloads

📄 Paper

DOI: https://doi.org/10.5281/zenodo.19152178

This repository contains the research work on Meta-Computing, a paradigm for self-optimizing distributed AI systems.

🔍 Abstract

This work introduces a Meta-Computing Architecture (MCA) that enables systems to:

• Observe runtime behavior
• Evaluate performance dynamically
• Optimize execution using feedback loops

Key contributions:

• Adaptive Meta-Optimization Algorithm (AMOA)
• Workload Complexity Index (WCI)
• Meta-Efficiency Score (MES)

🧠 Architecture Overview

The system consists of:

• Base Computation Layer
• Observation Layer
• Meta-Analysis Layer
• Meta-Control Layer

🔁 Feedback Loop

Continuous optimization using: Observation → Analysis → Decision → Adaptation

📊 Results Summary

Workload	Static Time	Meta Time
Low	40 min	36 min
Medium	120 min	95 min
High	300 min	220 min

📁 Repository Structure

• paper/ → Research paper
• diagrams/ → MCA & AMOA visuals
• implementation/ → Simulation code (future)
• results/ → Experimental outputs

🚀 Future Work

• Real-world deployment
• RL-based optimization
• Cloud-native implementation

👤 Author

Hemant Kumar Kushwaha
Assistant Professor, Haridwar University
ORCID: 0000-0002-1365-6063

⭐ Citation

If you use this work, please cite using the DOI.