Gancho Stoychev

Software engineer · Linux & AI infrastructure

Root since 2005.
Frontend since 2016.
Now: GPUs, containers and LLM inference in a home lab I break on purpose.
Real benchmarks, raw terminal output.

Two MSI GeForce RTX 5060 Ti cards stacked in the Z790 home lab server — The Z790 node: two RTX 5060 Ti cards that run the labs.

Latest posts

A container in 81 lines of Bash: what the kernel primitives cost on Ubuntu 24.04
Namespaces, cgroups v2, pivot_root and a veth pair, assembled by hand on an AI home lab: real outputs, real mistakes, and 124 ms against Docker’s 518 ms.