Modern distributed information systems increasingly require sophisticated automation appro- aches for task planning, resource allocation, and execution management. While AI agents based on large language models (LLMs) demonstrate significant potential for solving complex distributed computing tasks, fully autonomous solutions present substantial risks without proper human oversight and control mechanisms. This paper proposes a comprehensive hybrid multi-agent architecture that effectively combines automated planning and execution capabilities with strategically integrated human-in-the-loop (HITL) components for distributed intelligent systems. The proposed framework employs a hierarchical structure consisting of orchestrator and executor agents working in coordination with human specialists at critical decision points. The orchestrator agent manages high-level task decomposition and resource allocation across heterogeneous computing clusters, while executor agents handle local optimization and code generation for specific computational nodes. The hybrid nature of this architecture results from applying centralized orchestration at the task level while maintaining decentralized execution at the subtask and individual cluster level. Key innovations include the formal definition of agent interaction protocols, comprehensive toolsets for both orchestrator and executor agents, and multi-layered verify- cation mechanisms that combine automated static code analysis with mandatory human expert review. The HITL components are strategically positioned at critical junctions including task decomposition confirmation, resource allocation approval, and code verification before execution on real distributed infrastructure. This approach addresses the fundamental challenge of balancing automation efficiency with operational safety and control. The framework's practical applicability is demonstrated through clearly defined agent responsibilities, tool specifications, and interaction mechanisms that enable safe deployment in real-world distributed computing environments. This research contributes both practically and scientifically to the field by providing a structured approach to deploying AI-driven distributed computing solutions while maintaining necessary human oversight, formally characterizing hybrid hu- man-AI collaboration in multi-agent systems and establishing robust verification protocols that ensure operational safety without sacrificing computational efficiency.
- Al-Fraihat, D., Sharrab, Y., Al-Ghuwairi, A.-R., Alzabut, H., Beshara, M., & Algarni, A. (2024). Utilizing machine learning algorithms for task allocation in distributed agile software development. Heliyon, 10(21), e39926. doi: 10.1016/j.heliyon.2024.e39926
- Kankaniyage Don, R. T. P., Ravi, A., & Toxtli, C. (2024). Assessing the Task Management Capabilities of LLM- Powered Agents. doi:10.13140/RG.2.2.11776.85768
- Li, Z., Wu, W., Wang, Y., Xu, Y., Hunt, W., & Stein, S. (2025). HMCF: A Human-in-the-loop Multi-Robot Collaboration Framework Based on Large Language Models. doi: https://arxiv.org/abs/2505.00820
- Mao, H., Schwarzkopf, M., Venkatakrishnan, S. B., Meng, Z., & Alizadeh, M. (2019). Learning Scheduling Algorithms for Data Processing Clusters. doi: https://arxiv.org/abs/1810.01963
- Mediakov, O., & Khorkanin, M. (2025). Automated Planning in Intelligent Distributed Systems Using a Multi- Agent Approach Based on LLM. Visnyk of Vinnytsia Politechnical Institute, 179(2), 111–117. doi:10.31649/1997-9266-2025-179-2-111-117
- Sypherd, C., & Belle, V. (2024). Practical Considerations for Agentic LLM Systems. doi: https://arxiv.org/ abs/2412.04093
- Takerngsaksiri, W., Pasuksmit, J., Thongtanunam, P., Tantithamthavorn, C., Zhang, R., Jiang, F., ... Wu, M. (2024). Human-In-the-Loop Software Development Agents. doi:10.48550/ARXIV.2411.12924
- Zhang, W., Zeng, L., Xiao, Y., Li, Y., Cui, C., Zhao, Y., ... An, B. (2025). AgentOrchestra: A Hierarchical Multi- Agent Framework for General-Purpose Task Solving. doi:10.48550/ARXIV.2506.12508
- Zou, H. P., Huang, W.-C., Wu, Y., Miao, C., Li, D., Liu, A., ... Yu, P. S. (2025). A Call for Collaborative Intelligence: Why Human-Agent Systems Should Precede AI Autonomy. doi: https://arxiv.org/ abs/2506.09420