Multi-Site VPN Engineering and Network Security

Multi-site VPN topology, segmentation and hardening for organizations that have to keep operating through the bad day.

Context

Multi-site VPN engineering rarely starts on a clean sheet. The network already exists, which is the problem. It grew one device at a time, it is flat, the credentials are shared, and the only person who understood the address plan left. Meanwhile the organization now depends on it: cameras, servers, remote access for people who are not in the building, and sites in more than one country that all have to reach each other. The work here is the operator version of that problem, not the residential one: a documented address plan, segmentation with a reason behind every boundary, encrypted paths between sites, identity in front of internal services, and monitoring that finds a fault before a user reports it. Home and small business network work is a separate part of this company and lives on Network Design.

Scope

What this covers

  • 01Multi-site VPN engineering: WireGuard hub and spoke between a cloud hub and branch sites, with subnets segmented per site and peer handshakes monitored explicitly rather than assumed.
  • 02Network segmentation and hardening, replacing flat networks with a segmented architecture and a written address plan.
  • 03Identity and access in front of internal services: single sign-on with Microsoft Entra OIDC, Traefik forward-auth, and group-based role checks.
  • 04Routing and edge: custom OpenWRT routing servers, controller-based access point deployment, managed switch and PoE configuration.
  • 05Remote administration without exposure: tunneled access, scoped SSH, jump paths and least-privilege service accounts.
  • 06Network monitoring as an operational discipline: NetXMS, Prometheus node_exporter, reachability and resource workers, and alerting with a typed event taxonomy rather than an inbox of raw traps.
  • 07Vulnerability assessment and written reporting, including supply-chain audit of dependencies across an estate after a published CVE.
  • 08Penetration testing tooling and adversarial review of internal tools and third-party contractor software.
  • 09Hardware and software reverse engineering: protocol analysis on RF, keycard, USB and serial devices, logic analyzer level work, and undocumented device behavior established by measurement.
  • 10Carrier and circuit procurement support: requirements definition, bid coordination, technical due diligence and comparison reporting.
  • 11Documentation as a deliverable: topology diagrams, address plans, runbooks and change records that survive the engagement.

Record

Work behind this

Delivered work, stated as facts rather than claims. Programs and employers are named where naming them is what makes the record checkable. Clients are not.

R-01Replaced consumer-grade networking at a client site, an open WiFi network on a flat LAN, with a hardened multi-segment architecture: a custom OpenWRT routing server, correctly provisioned access points, and the controller running as a container on a cluster built for it.
R-02Multi-site VPN engineering delivered as a hub and spoke WireGuard mesh across a cloud hub and two branch sites, with subnets segmented per site and explicit peer-handshake monitoring.
R-03Internal services published behind Traefik v3 with Microsoft Entra OIDC single sign-on and a group checker enforcing role-based access, with a tunnel fronting one site.
R-04A production NetXMS monitoring platform standing at 18 nodes, live for ongoing network telemetry across a multi-site estate.
R-05A supply-chain vulnerability audit across an entire infrastructure estate following published CVEs on a widely used HTTP library, with every affected service remediated.
R-06Critical security vulnerabilities identified and reported in internal tools and in third-party contractor software at a former employer.
R-07A 4 Gbps fiber WAN procurement study: bids and technical due diligence coordinated across three carriers and an equipment vendor, with comparison reporting automated by an in-house agent.
R-08Sole ownership of network and IT infrastructure engineering for a US logistics company with international field operations, including direction of the in-house IT team and 24/7 on-call escalation.
R-09Public reverse engineering work, including hotel keycard system research with a Proxmark3 and protocol documentation for a bench power supply.

Questions

Questions technical buyers ask

How is this different from the network work on the services side of the site?

Audience and scale. Network Design is a home or a small business at one address, with a free on-site estimate and a price range on the page. This is an organization with multiple sites, an identity provider, staff who need remote access, and an obligation to document what it runs. Same engineer, different problem, and the deliverables look nothing alike.

Can you work on a network that is already carrying production traffic?

That is the normal case, and it sets the method rather than preventing the work. Discovery first, then a written plan with a change order per boundary, then cutovers in windows with a rollback path defined before anything moves. The risky pattern is not touching production, it is touching production without knowing what the current state actually is, which is why discovery is a deliverable rather than a formality.

What does a first engagement usually look like?

An assessment: the topology and address plan as they actually are rather than as documented, how sites reach each other, what is exposed, who can reach what and with which credentials, and where monitoring has blind spots. The deliverable is a diagram, an inventory and a ranked risk list. Remediation is quoted separately once both sides are looking at the same picture.

Start a technical conversation

Describe the system, the constraint and the deadline. An engineer answers.

Tools and platforms we work with: WireGuard and OpenWRT. Traefik v3 with Entra OIDC. UniFi controllers and access points, managed switches with PoE. NetXMS, Prometheus node_exporter, NUT. Debian, Ubuntu and Kali. Proxmark3, logic analyzers and protocol analysis tooling.