Blog

One in twelve packs of smoked salmon in German supermarkets has detectable listeria. And it's not from the fish.

tokens go brrrr

Our €5M funding announcement. Why we started UMH, the problem we're solving, and what's next.

A guide for LLMs (and humans) to write in my voice. Self-demonstrating.

How to properly use LLMs to prevent producing slop.

A Christmas Eve journey. Deutsche Bahn. 35 kilometers.

Featured on HN — 1164 points, 1008 comments

Multi-part video series demonstrating practical downsampling techniques including dead-band, swinging door trending, and report-by-exception.

Why most factory-floor LLM demos break apart in production - and how a generate, freeze, test workflow turns AI into reliable pipelines.

Most MQTT tutorials are either too basic or buried in protocol specifications. This explains what MQTT actually does and why it matters for manufacturing.

Writing technical articles is an art. This is wrong. It's a methodological process.

An analysis of cloud infrastructure costs, highlighting why we recommend Hetzner for trying out the UMH.

Featured on HN — 131 points, 138 comments

Why traditional manufacturing software fails and what we do instead.

How modern cloud-native technologies fit with the traditional automation pyramid.

Co-authored with Denis Gontcharov. Why we chose Grafana over traditional industrial visualization tools.

A critical look at OPC UA and its impact on the manufacturing industry.

What Unified Namespace actually is and why it matters.

The roles of OLTP and OLAP databases, and Lambda Architecture for managing large volumes of real-time and historical data.

Historians vs open-source databases in oil and gas, pharma, and chemicals. Compression, storage costs, compatibility.

Designing IT/OT infrastructure without MQTT, Kafka, and Kubernetes? We tried that and hit many problems.

Why we chose Flatcar for Industrial IoT. Our requirements and how we decided.

Technical comparison of Mosquitto, VerneMQ, EMQx, and HiveMQ. Which one to pick and why.

OSIsoft PI, Canary, InfluxDB, TimescaleDB, Ignition. Overview of historians vs time series databases.

How we handle data processing at UMH. Tools and techniques that work.

How an open-source tool is establishing itself in a highly competitive environment against billion-dollar companies.

Featured on HN — 91 points, 28 comments

TimescaleDB is better suited for the Industrial IoT than InfluxDB because it is stable, mature, and failure-resistant.