Infrastructure consulting, migrations and incident response. I work on the systems that are not allowed to fail: Proxmox clusters, production databases, Laravel and Next.js applications. Independent, based in Iași, working with clients in Romania and the UK.
I don’t deliver workarounds. I deliver the cause, the fix and the documentation.
Clusters that survive a node failing, not just a good day.
I move production systems without losing data or weekends.
The server is compromised now. First job is to stop data leaving.
Models running on your hardware. Data never leaves the network.
I write the application and stay accountable for it in production.
Alerts that mean something, and email that reaches the inbox.
Published anonymised, with no client names, addresses or system names. Each entry says what was broken, what I did, and what was left behind.
PostgreSQL · PostGIS · Laravel · RabbitMQ · Jenkins · KVM
Next.js · Docker · Oracle Cloud · iptables
AlmaLinux · cPanel / WHM · LiteSpeed · systemd
Next.js 15 · NestJS · MariaDB · Docker
Docker · Proxmox · Caddy · MariaDB
Ollama · LM Studio · Docker · Python · Home Assistant
Proxmox VE · Proxmox Backup Server · NFS · ZFS
Postfix · SPF · DKIM · DMARC
Laravel 11 · React · MySQL · Stripe · Docker
Thirty minutes, free, no questionnaire. If the problem isn’t mine to solve, I say so in that same conversation.
First I inventory what exists: services, backups, versions, single points of failure. I touch nothing in production until I know what I’m touching.
Fixed price for fixed scope, agreed before we start. Every intervention has a way back, tested before it is needed.
What changed, why, and how to fix it at three in the morning without me. Consultant lock-in is a defect, not a business model.
You learn what you have, what’s broken and what fails next.
Defined scope, fixed price, a delivery date.
Someone looks at your infrastructure every month.
Compromised server, production down, data lost. I respond the same day and I don’t ask for a signed contract before we stop the bleeding. No minimum hours.
Standard rate 150 RON/hour for work outside the packages. On-site travel billed separately. All prices exclude VAT — I am not VAT registered. Invoicing in EUR or GBP available for clients outside Romania.
Yes. Most of my clients have fewer than fifty staff and nobody dedicated to infrastructure. That is exactly where the gap between "it works" and "it worked" is widest.
No. The first stage is read-only: inventory, versions, backups, logs. Write access comes once I have a map of the system and a rollback plan.
That is why there is a restore-verified backup and a rollback plan before every intervention. I have moved dozens of production virtual machines without data loss, including databases holding decades of history.
Yes. The takeover audit exists precisely for the case where nobody remembers how the system is configured and the documentation left with the person who wrote it.
Yes, and it is one of the things I have done most often over the past year. Twenty-billion-parameter models run usably on a consumer GPU, inside your network, without a single token leaving the company.
Yes, as standard, before any access. The case studies on this site are published anonymised, with no client names, IP addresses or system names.
Same working day for ordinary enquiries. For an incident in progress, hours.
Thirty minutes, free. If the problem isn’t mine to solve, I tell you in the first conversation, and if I know the right person, I point you there.