What we do, specifically.

01 / Networks & NOC

The network, and the people who watch it.

Most of our network work is for wireless ISPs and multi-site operators: MikroTik and Ubiquiti estates with fibre, 60 GHz and licensed-band backhaul, OSPF areas, service tunnels per client, and a carrier handover at the end. We design and build, and we also sit on the other side as the NOC that notices.

• Design and build: PoPs, high-sites, last-mile radio and fibre
• Routing: OSPF areas and filters, BGP, BFD, queue trees that match the plan sold
• Overlay: WireGuard management meshes, Layer 2 service tunnels, SD-WAN
• CPE standards: configuration audits, a golden template, batched rollouts
• Configuration recovery when hardware dies and the export is old
• Path forensics: every router on a path checked and written up per event
• Read-only first: nothing changes on a live network without a file and a rollback

02 / Monitoring & reporting

Numbers an operator can forward.

We build and run LibreNMS estates, the dashboards that sit on monitoring-centre walls, and the reports that go to clients and carriers. The rule on all of it: a state needs two witnesses, carries a confidence, and links to the raw data. A monitoring system that cries wolf is switched off within a month.

• LibreNMS build-out: devices, groups, maps generated from topology, alerting
• Availability dashboards with client, NOC and admin roles
• An hourly report with an issue ledger and cascade root cause
• Monthly client availability reports in the carrier’s own format
• NetFlow and IPFIX collection, device naming from what the network knows
• Dispatch to WhatsApp and voice, one message per event
• Written measurement rules, printed on the page they govern

03 / Hosting & email

A server we patch, back up and answer for.

Terawiz started here. Client websites and mail run on our own Debian server in South Africa with a panel clients can use themselves, DNS on Cloudflare, and the mail authentication that gets your messages into inboxes. When a host is compromised, ours or anyone else’s, we do the containment and write the report.

• Websites, mailboxes and DNS for small and mid-sized businesses
• Migrations from shared hosts, mailbox by mailbox, with a per-domain report
• DKIM, SPF and DMARC per domain; hybrid relay alongside Microsoft 365
• TLS certificates, backups, hardening, rate limits on the things that get hammered
• Incident response on compromised hosts: contain, preserve evidence, report
• Invoices and statements from one address, billing@, and nowhere else

04 / Software & apps

Tools that come out of running something.

Almost everything we write starts as a tool we needed on a live system: a report, a dashboard, a restore script, a collector. It gets kept, documented and reused. We also ship consumer apps, and have since 2011. Where a language model helps, it drafts and a person decides; the data never depends on it.

• Operational dashboards and small APIs in Python and Node
• Integrations: RouterOS and UISP APIs, LibreNMS, WhatsApp, Microsoft 365
• iOS in Swift and SwiftUI, Android in Kotlin, published under our own name
• Language-model assistance with a human in the loop, local models where data must stay put
• Digital signage, WebXR showrooms, voice dispatch over SIP
• Repos, deploy scripts and a README for everything handed over