I build technology myself and document the journey.

lon3s is my personal build journal for technology projects that I plan, build, experiment with and document step by step. The focus is on hardware, Linux, homelabs, Proxmox, local AI and self-hosted systems.

It’s not just about showing a working result at the end. I want to document the journey as well: What hardware was used? Why did I choose a particular solution? What worked, what didn’t, and which problems had to be solved along the way?

For me, those experiences are part of the project. That’s why I document not only the finished setup, but also the mistakes, detours and decisions along the way. The result is a guide that doesn’t just show what was built, but also explains how it came together — so others can rebuild the project themselves and learn from the experience.

For a long time it was assumed: for serious AI applications, you cannot get around the cloud. The required computing power was too expensive or too complicated to run on your own.

That is no longer true. Open-weight models are now powerful enough to run on your own hardware, without a cloud subscription, without your data ending up somewhere else. That is exactly what I want to show: how to build it yourself, step by step, with real hardware and real results.

I am Senol.

My professional background is in metalworking, but technology has always been a big part of what I do outside of work.

I build my own computer systems, run a homelab, work with Linux and Proxmox, experiment with local LLMs and explore ways to run more technology myself instead of relying entirely on the cloud.

I learn by building things, breaking them and figuring out how to make them work.

Server, workstation and a mini PC, every device has its own role.

Server (Proxmox, running 24/7): Gigabyte AMD A520 So.AM4 DDR4 Mini-ITX · AMD Ryzen 3400G · 16 GB RAM · 2 TB SSD + 8 TB HDD · RTX 5060 Ti 16 GB VRAM. Deliberately tuned for efficiency, idles at around 36 watts.

Workstation: ASRock X670E Taichi · Ryzen 7800X3D · 32 GB RAM · 1 TB SSD · 2× RTX 3090 (24 GB VRAM each). For everything that needs more power, above all local AI models.

Mini PC: ASRock Deskmini X300 · Ryzen 5700G · 32 GB RAM · 1 TB SSD. For testing and everything that should run in isolation.