🌴koko
Enrique Eisenkolbl
Open to opportunities Enrique Eisenkolbl Β· Site Reliability Engineer Β· San Lorenzo, Paraguay

I make 3am boring.

Currently Β· ICDF Taiwan technical cooperation, πŸ‡΅πŸ‡Ύ side
Reliability and observability for Paraguay's national health information system β€” γ€ŒProyecto HIS」
90+ hospitals and clinics monitored Β· Elastic alerting + case tracking Β· incident response

6+ years keeping production systems alive: hosting for 230+ clients, then fintech, now national healthcare. At Proyecto HIS the incident log used to be a shared Excel sheet somebody remembered to update β€” I replaced it with ElastAlert2 and Kibana cases, so every event has an owner and a trail. I also write the backup checks that catch the failures reporting green while doing nothing. Terraform, Ansible, Go and Bash underneath it all. Bilingual en/es. Open to remote SRE and platform reliability roles.

See what I keep running Email me β†—
🐱 kitty fish: ~/EE · enrique@bashkoko
fish: ~/EE
htop
ssh his-bastion
1[] 42%
2[] 69%
3[] 15%
4[] 34%
Mem[]18.2G/32.0G
Swp[] 0.3G/8.0G
Tasks: 127, 89 thr; 2 running
Load avg: 1.24 0.98 0.87
Uptime: 5 days, 14:22:07
Host: his-elk-01 (LXC)
OS: Debian GNU/Linux 12
Kernel: 6.1.0-21-amd64
PIDUSERPRINIRESSCPU%MEM%Command
1842elastic2006.2GS31.419.4java [elasticsearch]
2103logstash2001.8GS8.25.6java [logstash]
3721kibana200512MS4.11.6node /usr/share/kibana/bin/kibana
842postgres200312MS2.10.9postgres: his_db checkpointer
1244root200444MS1.41.4dockerd
3901enrique20012MS0.80.0fish
411root2004MS0.20.0nginx: worker process
382root2004MS0.00.0sshd: /usr/sbin/sshd -D
1root20012MS0.00.0/sbin/init
2847enrique20012MR0.00.0htop
F1Help  F3Search  F5Tree  F6SortBy  F9Kill  F10Quit
enrique@bashkoko ~/EE git:(main) ❯ ssh his-bastion
Warning: Permanently added 'his-bastion' (ED25519) to the list of known hosts.
Welcome to Ubuntu 22.04.4 LTS (GNU/Linux 5.15.0-112-generic x86_64)
Last login: Tue Jun 3 22:15:11 2026 from 192.168.10.1
enrique@his-bastion ~ ❯

02 Β· experienceExperience

πŸ“‹ Work History experience.log
Sep 2025 to Present
● active

DevOps Engineer Β· ICDF Taiwan

πŸ‡ΉπŸ‡Ό πŸ‡΅πŸ‡Ύ  Taiwan–Paraguay technical cooperation Β· backend infra for the national health information system Β· γ€ŒProyecto HIS」

I own the reliability layer for a healthcare platform used across 90+ public hospitals and clinics. I inherited an incomplete ELK 8.14 deployment and a separate Prometheus/Grafana stack and consolidated both into one production-ready ELK 9.x cluster ingesting ~120 GB/day. On top of it: ElastAlert2 querying Elasticsearch directly for faster detection, Kibana case automation that replaced a hand-maintained Excel incident log, uptime monitors as code, backup monitoring built on Elasticsearch Transforms that catches empty-but-green failures, and a weekly health report that ministry leadership actually reads. Monitored through hospital hours, with me as the escalation after. Everything reproducible with Ansible and Terraform β€” a cluster rebuild is hours, not weeks.

ELK StackObservabilityAlertingIncident ResponsePostgreSQLAnsibleTerraformGoBashPython
πŸ—” open workspace Β· 20 projects
Jul 2024 to Sep 2025

Infrastructure Analyst Β· Negofin S.A.E.C.A.

Ran on-prem infrastructure for a fintech: 120+ VMs across an 8-node VMware cluster, 4 NAS devices and 20 switches. I ran the DR environment and built the test suite that validated updates before they reached production, automated Ubuntu provisioning with cloud-init, centralized syslog from switches, servers and FortiGate firewalls, and kept Veeam backups replicated off-box so a bad night had somewhere to fall back to.

VMware vSphereDisaster RecoveryDockerVeeamcloud-initsyslogVLANs
πŸ—” open workspace Β· 4 projects
Dec 2019 to Jul 2024

Systems Administrator Β· Saiternet

Supported 230+ hosting clients over nearly five years β€” cPanel/WHMCS, mail, DNS, SSL β€” plus the blade servers, pfSense firewalls and MikroTik routers underneath. I wrote the FAQs and troubleshooting guides that cut repeat ticket volume. Five years of being the person clients only notice when nothing breaks.

cPanelSMTPMikrotikHostingSSL/TLS
πŸ—” open workspace Β· 4 projects

03 Β· workSelected Work

πŸ—‚ Projects ~/homelab/projects/
01self-hosted SSO

Zero-Trust Gateway

Every homelab service β€” including this site β€” sits behind one default-deny edge: nginx terminates TLS, Authelia enforces TOTP 2FA, and a Cloudflare Tunnel means my home IP never appears in DNS. 8 services, one login, zero open ports.

nginxAutheliaCloudflareLet's Encrypt
02infrastructure-as-code

Proxmox Homelab as Code

~15 LXC containers described in Terraform and configured by Ansible, with scheduled backups and static addressing. I can rebuild the entire fleet from a fresh Proxmox install with two commands β€” and I have, which is the only reason I trust the claim.

ProxmoxTerraformAnsibleLXC
03metrics + logs

Observability Stack

A dedicated node running the same ELK setup I operate at work, pointed at my own fleet. It's where alert rules get tested against infrastructure I can afford to break, before they ever fire at a hospital.

ElasticsearchLogstashKibana
04private inference

Self-Hosted LLM

Ollama on local hardware behind a private chat UI. No prompts leave the network and no usage bill arrives β€” the same constraint I design for at work with patient data.

OllamaLLMAPI
05isolated workloads

Game Server Fleet

Terraria, Unreal Tournament, Abiotic Factor for friends. Each game lives in its own isolated LXC with scheduled snapshots, so a modded server gone wrong can't take the rest of the fleet with it and a wipe takes one command to undo.

LXCsystemdsnapshots
06self-hosted media

Media Platform

Jellyfin plus the *arr stack, each piece in its own LXC, deployed by the same Ansible/Terraform pipeline as everything else. The point isn't the movies β€” it's that the homelab is one fleet, managed one way.

Jellyfin*arrLXCDocker

04 Β· stackStack & Capabilities

βš™ Tools capabilities.conf

Systems

  • Linux
  • Windows Server
  • cPanel
  • CloudLinux

Virtualization

  • Proxmox
  • Docker
  • VMware
  • Hyper-V
  • Kubernetes

Networking

  • VLANs
  • DNS / DHCP
  • VPN
  • Mikrotik
  • Routing

Security

  • OPNsense
  • Fortinet
  • UFW / iptables
  • Zero-Trust

IaC & CI/CD

  • Terraform
  • Ansible
  • GitLab CI/CD
  • GitHub Actions
  • Bash / Go / Python

Observability

  • Prometheus
  • Grafana
  • ELK Stack
  • Observium
  • Veeam

AI & LLM

  • Ollama
  • Claude
  • ChatGPT
  • Local Inference
  • Prompt Engineering
  • LLM Automation

05 Β· aboutAbout

πŸ‘€ About about.md
C1   ENRIQUE #2026
Enrique at his homelab
/san-lorenzo CARD_01

I write the Bash and Go that watch things β€” backup monitors, ELK ingest pipelines, alert rules, and the Ansible that puts them everywhere. I use AI to move faster, but I review every line that ships to prod. The architecture is mine; the keystrokes are cheap.

I'm the person teams call when the cluster is on fire at 2am and the runbook says "check Enrique's stuff first." I lean toward boring tech, clean rollbacks, and writing the SOP I wish I'd had the first time. I work in English or Spanish, async or sync, and I'll always tell you when I don't know something.

πŸ“œ Studying: CKA (Certified Kubernetes Administrator) β€” target Q4 2026 Β· AWS SAA queued behind it.

πŸ“ Postmortem-style writeup Β· the master-slave / ECK shared-fate question

On the HIS observability stack, the proposed path was: master on host A, two more hosts for workers, more cost, more hardware. The slaves couldn't survive master loss, so the team put two on every host β€” a shared-fate anti-pattern dressed up as HA.

The counter-proposal I documented: ECK on a single control-plane node, with Elasticsearch data nodes as StatefulSets, HA coming from etcd quorum rather than physical host separation. Cost-neutral vs the current setup, no shared-fate coupling, recoverable from etcd snapshots. Stakeholders chose the familiar path. I kept the writeup β€” it's a good exam answer for "what would you do differently."

// this site is part of my Homelab.
built with Astro Β· provisioned via Terraform + Ansible
served behind the same zero-trust gateway as everything else.

06 Β· contact

Let's keep something important running.

Open to Site Reliability, platform, and infrastructure roles β€” remote or relocation. Bilingual (es/en), at home on distributed international teams.

Two ways to reach me β€” pick whichever is fastest for you.

Email Β· eeisenkolbl@bashkoko.com See the code β†—