Operationalizing LLMs in the Network Ecosystem

• Developing a novel approach to automated network topology discovery using large language models (LLMs)
• Leveraging LLMs to communicate with routers via NETCONF
• Generating structured JSON outputs to identify neighboring routers and recursively expand the topology via REST APIs
• Combining LLM reasoning with protocol-based inputs to dynamically visualize complex, multi-layered network architectures

• Understanding how to host LLMs using edge compute and integrate them into generative AI systems
• Exploring the role of reinforcement learning and fine-tuning to customize models for industry-wide and agentic AI use cases
• Reviewing existing LLM hosting platforms—open-source and commercial—with emphasis on low-latency needs and forward-looking hosting strategies
• Evaluating experience of varying hosting models as well future and proposed solutions

Technical paper presentation.

• Identifying risks and inefficiencies in end-to-end order fulfillment due to fragmented tools, manual hand-offs, and complex product/service configurations
• Demonstrating how LLM-based diagnostic analytics can identify root causes and recommend resolutions using enterprise data from ERP, CRM, WMS, and other systems
• Enhancing order accuracy, site readiness, and inventory availability through prescriptive AI, including automated bill of materials generation
• Emphasizing the role of Agentic AI and human-in-the-loop feedback to ensure continuous process improvement, data security, and compliance