Skip to content

8.98°N 79.52°W · Panama City· UTC−5

Javier Kerr

Infrastructure that stays up.Agents that watch it.

I'm an infrastructure engineer and independent consultant based in Panama. I've spent 20+ years keeping servers, networks and clouds running; 7 of them in business continuity, 3 working inside Veeam. Today I build AI agents with n8n and the Claude API that watch over that infrastructure and take repetitive work off your team.

Get in touchSee my work

$ jk --status

role
independent consultant
focus
infrastructure · backup/DR · AI agents
base
Panama · remote across Latin America
stack
Veeam · Proxmox · n8n · Claude API
status
taking on projects

$

01 Selected work

What have I built?

Everything I recommend runs in my own operation first. These are four systems I designed, built and use.

01 / 04

Orion

A SysOps agent that acts instead of sending one more alert email.

own lab · in use

  • Next.js 14
  • NestJS
  • PostgreSQL 16
  • Redis 7
  • BullMQ
  • Claude API
  • n8n
  • Proxmox
Context
Traditional monitoring tells you something broke, and that is where its job ends. Orion reads infrastructure telemetry, cross-checks backup state against security alerts, and executes the response.
Decisions
  1. 01Agents on the Claude API cross-check Veeam against Wazuh before escalating anything.
  2. 02n8n as the execution engine and a Telegram bot as the command console.
  3. 03Docker Compose on Proxmox, ~120 MB Alpine images and separate internal and public networks.
Outcome
It runs on my own infrastructure. It is the test bench for everything I later propose to a client.

02 / 04

Heimdall AI · Hofund

An AI vCISO for MSPs: it analyzes, proposes and waits for your approval.

own product · in development

  • Node.js
  • Express
  • React
  • Vite
  • PostgreSQL
  • Claude API
  • n8n
  • Docker
Context
A small MSP can't afford a CISO, but its clients still get attacked. Hofund ingests raw logs from Proxmox, Docker, Veeam and SSH and turns them into alerts with severity and business impact.
Decisions
  1. 01Human approval required: Claude proposes the remediation script and never runs it on its own.
  2. 02Domain health check (SPF, DMARC, MX, SSL) with a 0-100 score and a monthly roadmap.
  3. 03Moved in August 2026 from GCP/Firebase to self-hosted Docker and PostgreSQL: client data stays home.
Outcome
In development on my own stack. It is the Nibel Protect thesis — security and continuity as a service — packaged so an MSP can offer it.

03 / 04

Nibel Protect help desk

Every support email becomes a ticket, without anyone copying it by hand.

in production · Oct 2026

See Nibel Protect on nibel.tech

  • Dolibarr
  • n8n
  • PostgreSQL
  • Docker
  • PHP
Context
Support arrived as loose email and depended on memory. I centralized it in Dolibarr Tickets as the source of truth. The n8n workflows that classify, draft replies and chase what is pending are built and switch on in stages.
Decisions
  1. 01Staged rollout, each stage approved and backed by a written rollback plan.
  2. 02AI stays off until it is defined which data may leave. Privacy first, under Panama’s Law 81.
  3. 03Nothing reaches a client without human review: n8n drafts, I send.
Outcome
Email → ticket in production since October 16, 2026. The mailbox is collected every 5 minutes, and no support email depends on someone remembering it anymore.

04 / 04

AI OS · single screen

My whole operation on one screen, with the data in one place.

own operation

  • Docker Compose
  • n8n
  • PostgreSQL
  • Grafana
  • Paperless-ngx
  • DocuSeal
Context
Finances in a spreadsheet, hosting in another panel, social in a third. I built a container stack that brings it together and answers one question: what needs my attention today.
Decisions
  1. 01Data lives in PostgreSQL. n8n writes; Grafana and the screen only read.
  2. 02Local first; it moves to a VPS by changing only the .env file.
  3. 03Paperless-ngx for documents and DocuSeal for signatures join the same stack.
Outcome
One entry screen for the day and Grafana for the detail, without opening five tools.

02 Work with me

How can I help?

I work solo, with my own lab behind me. You talk directly to the person who designs and delivers.

  • 01

    Backup and continuity

    You have backups. Good. When did you last restore something? Immutable repositories, real restore tests and a recovery plan someone has already rehearsed.

    Veeam · Datto · Nibel Protect

  • 02

    Infrastructure and cloud

    Clusters, migrations and hybrid environments that hold up on a bad day. From the hypervisor to Microsoft 365 and Entra ID.

    Proxmox · VMware · KVM · M365

  • 03

    AI automation

    Workflows and agents connected to your systems, with human approval where it matters. They take the copy-and-paste hours off your team.

    n8n · Claude API · ConsultorIA

  • 04

    Presales and delivery

    For integrators and consultancies that sell more than they can deliver: solution architecture, SOWs and hands-on implementation.

    Architecture · SOW · Technical PM

Two ways to start: an assessment with fixed scope and price, like the Diagnóstico 360 for backup and AI, or a monthly retainer. Before any proposal, a short conversation at no cost.

Tell me about it

03 Writing

What do I write about?

Notes on backup, infrastructure, patching and automation. Every figure links to its source.

View all
  1. Cloudflare unifica su observabilidad: 8 mejoras claveOct 3, 2026 · Cloud & Observabilidad · in Spanish
  2. BigDiskBuster: el zero-day que deja a Windows Defender sin actualizacionesSep 26, 2026 · Ciberseguridad & Parcheo · in Spanish
  3. Veeam ya cubre el 95% de los hipervisores: qué trae la v13.1Sep 16, 2026 · Backup & Virtualización · in Spanish
  4. La nube no es infalible: prepárate para más caídas en 2026Sep 15, 2026 · Continuidad & Nube · in Spanish
  5. Proxmox VE 9.2 ya balancea el cluster solo: qué cambiaSep 14, 2026 · Infraestructura & Virtualización · in Spanish

04 About

Who's behind this?

I've seen backups fail in production. That's why I don't call data protected until I've restored it. I've worked as a systems engineer for more than 20 years; the last 7 in data protection and business continuity, 3 of them working directly for Veeam, not as an outside integrator.

Then I added the missing layer: AI automation. I use it for the repetitive work — checking, cross-referencing, alerting — so your team stops firefighting and a person still makes the call.

I build my own tools and use them before recommending them: the agent that watches my lab, the help desk, the dashboard where I see my operation. Since October 2026 I work full time as an independent consultant, almost always remotely.

years in infrastructure
20+
years in business continuity
7+
years inside Veeam
3
articles published
12

What I use every day

  • Veeam
  • Proxmox
  • VMware
  • Microsoft 365
  • Entra ID
  • Docker
  • n8n
  • Claude API
  • Wazuh
  • PostgreSQL

05 Questions

What do people usually ask?

Who is Javier Kerr?

Javier Kerr is an infrastructure engineer and independent consultant based in Panama City. He has 20+ years in servers, networks and cloud, 7 in business continuity and 3 working for Veeam. Today he combines backup and recovery with automation and AI applied to IT operations.

Do you only work in Panama?

No. Most of my services are fully remote, so I work with companies in Panama and across Latin America. On-site visits are arranged when a project needs them.

Do you work with small businesses?

Yes. A small business with a single server can lose all its data in one night too. The design fits the real size of your operation and your budget, not the other way around.

How is pricing structured?

There are two models: a project with a defined scope and deliverable, or a monthly retainer for infrastructure that cannot fail. We talk through your case first, free of charge, and then I send you a clear proposal.

What if I already have an IT team?

Even better. I work as an extension of your team on the specialized parts (backup, recovery, virtualization, automation) and leave the knowledge documented so your people become more autonomous.

06 Contact

Shall we talk about your infrastructure?

Tell me in two lines what worries you. The reply comes from me, not from an agent.

jk@javierkerr.com

It's UTC−5 · Panama City