15+ years in SD-WAN and SASE. Lately I've been building an approval-gated AI orchestration agent that connects to multivendor network infrastructure.

Pre-Sales Engineer, SASE & Security · HPE
15+ years of experience in cybersecurity, SD-WAN, and network services — with a growing focus on how AI fits into network operations.
I am a Network and Security Infrastructure Engineer with 15+ years of experience spanning hands-on network engineering and pre-sales. I started my career designing and operating large-scale networks directly — including a global network supporting roughly 1,500 locations and 18,000+ users — before moving into pre-sales, where I've spent the past several years helping customers and channel partners design, demonstrate, and implement SD-WAN and SASE solutions.
More recently, I've been building the Universal Network Director — an approval-gated AI orchestration agent that governs MCP servers across multiple network vendors — HPE Aruba EdgeConnect (REST), Juniper (NETCONF), Proxmox (REST), and Cisco (CLI over SSH) — using the Anthropic API to let AI systems query and act on real infrastructure. Read-only requests pass through automatically; anything that changes live configuration requires my explicit approval before it executes. I'm actively expanding the integration set with MCP servers for Juniper Mist and Juniper Security Director Cloud, both REST-based. I build with Claude Code, applying my own network engineering background to the protocol integrations, safety design, and approval workflow — not a claim to have solved this, just genuine curiosity about where AI fits into network operations.
Networks have always been multivendor by necessity — no single platform covers everything. I think AI's real value here isn't replacing that reality, it's finally giving us a consistent way to see and operate across it, as long as the humans stay in control of anything that actually changes production infrastructure.
A few things I think about when working on this.
Building an AI orchestration layer for multivendor network infrastructure — one integration at a time.
The orchestration agent at the center of the whole project — it sits in front of my vendor MCP servers and uses the Anthropic API to let AI systems query and act on real network infrastructure across HPE, Juniper, Proxmox, and Cisco. Read-only requests pass straight through; anything that would change live configuration is held for my explicit approval before it executes. Built with Claude Code, applying my own network engineering background to the protocol integrations, safety design, and approval workflow.
MCP server for HPE Aruba EdgeConnect Orchestrator over its REST API — exposes appliance status, interface and tunnel state, and WAN bandwidth data to AI systems, with read/write separation and secret redaction.
MCP server for Juniper devices running Junos, communicating over NETCONF — lets AI systems read device state and configuration, with changes routed through the Universal Network Director's approval workflow.
MCP server for Cisco IOS-XE devices over CLI/SSH — surfaces device state to AI systems and passes any configuration changes through the approval-gated workflow rather than executing them directly.
MCP server for Proxmox VE over its REST API — gives AI systems visibility into the virtualization environment behind the lab, with the same read-through / approval-gated-write model as the other integrations.
Extending the director to Juniper Mist's REST API — in progress.
Extending the director to Juniper Security Director Cloud's REST API — in progress.
A track record spanning hands-on network engineering, pre-sales, and channel enablement.
A broad technical foundation combined with strong business and communication skills.
Whether you're looking to discuss a network challenge, explore a collaboration, or simply connect — I'd love to hear from you.