AI systems
Agents, evaluation loops, model boundaries, and useful automation.
wolkengrube.de
Turning data into insight. Building systems that work.
Short intro
I write about AI-assisted engineering, data platforms, infrastructure, and the practical architecture decisions that happen between idea, notebook, pipeline, cluster, and production.
Current themes
Agents, evaluation loops, model boundaries, and useful automation.
Working with coding agents, review habits, and human-in-the-loop flow.
Workflows, Kubernetes patterns, deployment paths, and observability.
Lakehouse workflows, notebooks, jobs, governance, and platform habits.
From raw signals to governed datasets, usable models, and calm operations.
Small tests, notes, migrations, and repeatable discoveries in public.
Tiny experiments
Short field notes, architecture sketches, and small experiment logs are being shaped here. The goal is not polished certainty, but useful traces.