Websites and server infrastructure that actually work.
I build advanced websites and self-hosted server infrastructure — from a single business site to a cluster of devices unified into one secure fleet.
Services
Full range: from the site itself, through the server it runs on, to security and automation.
Advanced websites & web stores
Custom-built WordPress themes and integrations tailored to your business — not an off-the-shelf template. Fast, responsive, optimised for SEO and conversion.
HomeLab & self-hosted server infrastructure
Server environments built from scratch: Docker, reverse proxy, databases, backups, monitoring. From a single box to a cluster of several devices working as one system.
Local AI instead of the cloud
Deploying language and generative models on your own hardware — full data control, no per-query licensing, GDPR-friendly by design.
Security & monitoring
Security audits, server hardening, SIEM-based monitoring, automated attack detection and blocking, end-to-end encrypted backups.
Automation & custom dashboards
Automation agents for repetitive work (n8n), custom-built management dashboards — the same kind I use every day to run my own server fleet.
My proof, not just a promise
Instead of just describing skills — I show the infrastructure I designed, built and run every day: four devices unified into one server fleet. This site itself runs on a separate, dedicated VPS — the fleet is my private proving ground and proof of skill, not client hosting.
Illustrative visualisation — real monitoring (Grafana + Prometheus + Wazuh) runs 24/7 inside the fleet.
Reverse proxy & auth gateway for the whole fleet, media server and photo backup, network-level DNS filtering, ~16.5 TB of storage.
Shared database for every service, internal wiki, fleet-wide monitoring (metrics + alerts), encrypted backups, invoicing.
Security analysis & operations environment: fleet-wide SIEM aggregation, AI-assisted document analysis, honeypots, OSINT reconnaissance — 100+ tools.
37 local language models (including Polish-tuned ones) on dedicated GPU, image & 3D model generation, speech transcription, streaming — zero data sent to the cloud.
37 local AI models running on owned hardware — roughly 130 GB of models total, no data ever leaves to third-party clouds.
Four independent devices unified into one encrypted mesh network — key-only access, zero passwords exposed on the network.
Custom MCP servers on every node — AI agents can operate the fleet directly. Real agentic integration, not a single API wrapper.
Automated work modes — the infrastructure starts and stops services on its own (Idle / AI+Work / Desktop / Streaming) based on real usage.
SIEM (Wazuh) aggregating security events from all four nodes in one place, 24/7.
Custom IoT hardware (ESP32 Cardputer, LoRa mesh) integrated into the automation — security here means real hardware too, not just code.
What you're seeing is running code
This node network and the panel next to it are computed live in your browser — no external libraries. A small sample of what modern web tech can do.
What I use
Proven, production-grade technology — the same stack that runs my own fleet every day.
How working with me looks
Discovery call
We talk through your goal, budget and real needs — no selling you things you don't need.
Design & architecture
I design the solution: a site mockup or a server architecture, choosing technology that fits your actual scale.
Build & deploy
I build, test and deploy — with automatic SSL, backups and monitoring configured from day one.
Monitoring & support
After launch, the infrastructure is watched 24/7. I stay available for support and further development.
Questions I hear most often
By default, no. I prefer infrastructure I fully control — my own servers, my own AI models, my own backups. If a project genuinely needs a third-party service (e.g. payments), we agree on that explicitly before starting.
A simple business site: usually 1–2 weeks. A store or more complex server infrastructure: 3–6 weeks depending on scope. I always give a realistic timeline before starting, not after.
No. Every deployment includes monitoring, automated backups and my ongoing support — the same approach I use every day to run my own four-node server fleet.
Sometimes a cloud API is genuinely the best choice, and I use them too — but when data privacy, cost at scale, or offline operation matter, local models on owned GPU hardware are a real, battle-tested alternative.
Have a project that needs solid foundations?
A website, a store, a server, or full infrastructure — let's talk before you pick an off-the-shelf solution you'll need to rebuild anyway.
Get in touchHave a project in mind?
Tell me in a few sentences what you need — a website, a store, a server, or a full infrastructure build. I reply personally.
kontakt@lechovia.pl