About.md

I build things to understand them. Sometimes they work.
Sometimes they teach me something better than working would have.

I'm Christian T. Drieling — engineer, tinkerer, and collector of more side projects than I can name in one sitting. Based in Oldenburg, Germany.

By day I work with the kind of infrastructure people only notice when it breaks — virtual desktops, Microsoft cloud, identity, the parts of IT where downtime ends up in someone's calendar at 7am. By night I run a homelab where reliability is optional and learning is mandatory. The contrast keeps both honest.

This blog is where I write down what happened: the experiments, the infrastructure decisions, the AI rabbit holes I fell into at 11pm on a Tuesday. No polished theory — just applied curiosity. Try something, see what breaks, figure out why.

It started, slightly embarrassingly, because I kept opening new tabs to write things down anyway. At some point, creating this blog felt less ridiculous than another folder of experiment-final-FINAL-v3.md files.

What this is

Applied Curiosity is a personal lab journal. I write about things I've actually tried, built, or broken — with enough detail that future-me (and maybe you) can reproduce or avoid it.

A few recent rabbit holes:

  • Grounding Gradient — built a pipeline to test whether two LLMs drift into mutual hallucination without external anchors. The hypothesis was confident. The result was not.
  • Atlas — measured whether persistent memory makes a model more agreeable as the context grows. Turns out memory amplifies whatever the prompt rewards — in both directions.
  • LLM Telephone — push a text through ten summarization rounds across different modelsand see what makes it to the last hop. Currently running. The bet: not all models are equally careful messengers.

Topics I keep coming back to:

  • AI that runs on hardware you own
  • Self-hosted infrastructure and the lessons it keeps teaching me
  • LLMs in practice — what works outside the demo videos
  • Automation, the home lab, and the small joy of it works on my machine

What this is not

  • A tutorial site. Things break mid-post and I'll tell you.
  • Thought leadership. I'm figuring this out as I go.
  • Up to date. I write when something is worth writing about.
  • Vendor-neutral. I use what works for me, not what's popular.
  • A substitute for the docs. Read those too.

The stack

Local-first. Cloud where it makes sense. This list ages fast — assume it has drifted by the time you're reading.

Hardware

MacBook Pro M4 Max — 128 GB unified memory
Mac Studio M3 Ultra — 96 GB, dedicated to local inference

Local inference LM Studio · oMLX · Ollama on Proxmox

Models I run regularly

  • Text: Qwen3.6-35B-A3B · Gemma4 and friends
  • Voice (TTS): Qwen3-TTS · Voxtral-4B
  • Speech (STT): Parakeet-TDT
  • Embeddings & rerank: jina-embeddings-v5 + jina-reranker-v3

Agents & automation Hermes Agent · n8n · Open WebUI Pipelines and a collection of MCPs

Search & RAG Perplexica · Apache Tika · and the Jina stack

Cloud fallback — for when the model needs to be smart and fast.

Uptime: mostly.

Get in touch

Code lives at github.com/cdrieling. Mail goes to blog@apanthos.com — corrections, ideas, or "this broke in an interesting way" are all welcome.