Jaigouk.Kim

How a series of silent freezes during gpumod mode switches led me through CUDA pinned-memory debugging, four layers of defense, and ultimately to Qwen3.6-35B-A3B MTP with preserve_thinking as the new Hermes-agent default.

How I benchmarked Qwen3.6, Gemma 4, and multiple quantizations on an RTX 4090 to find the best local model for Hermes Agent — and how gpumod made the whole process manageable.

A bash function to manage Claude Code teammate sessions with tmux across multiple projects.

Programmer's point of view on OpenAI Codex. thinking about 5 years from now.

there is a way to build docker image for rust

hosting ghost blog via github static pages.

Deploying harbor arm64 registry to raspberry pi k3s cluster

In these days, I stay at home most of the time and still I wish to change

Istio service mesh on k3s rpi 4 cluster

> Thinking of yourself as a separate entity can reduce anxiety, while also kicking

no server. but network.

in 90's, mp3 came up. and then lots of mp3 players showed up in the market. but

personal experience about spreading openfaas at work

Nats vs Kafka vs etc. comparing message queues

Do you need a free kubernetes env for playing with openfaas?

I wanted to play with serverless architecture but I didn't want to spend money

Motivation

I moved to Berlin 1 month ago.

(Deis + Kubernetes + AWS) for the win

Learn once, write anywhere

the power shift from platforms to people would happen anytime soon.

Stack : deis, kubernetes, gitlab, node.js, ruby, elixir, react + redux

The ultimate glue for microservices