Raphael Wieland · WielandTech
I build software and keep it running
Kubernetes platforms, Django apps, data pipelines and LLM integrations. I write the code, run it on a cluster in my closet, and I'm the one who gets the alert when it breaks. Everything below is live.
- Live from the homelab
- The cluster this site runs on, right now.
- Availability · 30 days
- —
- Latency · p99
- —
- Requests / sec
- —
- Errors · 5xx
- —
- Saturation · CPU
- —
- Deployments healthy
- —
- Pods running
- —
- Nodes ready
- —
- Postgres data
- —
Loading live stats…How it's built →
Work
01 / Shipped and running
WARN Index
US layoff notices from 47 states and DC in one searchable place. When a state's site changes and a scraper breaks, an agent reproduces it and opens the fix as a pull request.
34scraper breaks fixed by reviewed PR
Live
- Python
- Playwright
- PostgreSQL
K3s homelab platform
A small production platform in a closet. GitOps with Flux, a review app for every pull request, managed Postgres, and full observability.
1,300merged pull requests
Live
- Kubernetes
- K3s
- Flux
Flock camera map
A map of 336,000 Flock Safety surveillance devices and license plate readers that doesn't track the people looking at it.
336Kdevices mapped
Live
- MapLibre
- PMTiles
- tippecanoe
r12d
Privacy-first QR stickers, short links, a photo inventory and lost-and-found for households and small businesses.
1stored row per code per day, not per scan
Live
- Django
- PostgreSQL
- S3
Restroom QR ad network
Venues put QR stickers in their restrooms, patrons scan them, and ad revenue is split across venues by traffic.
10ratings before a venue ranks
Live
- Django
- PostgreSQL
- ntfy
Services
02 / What I can build for you
01
Kubernetes & GitOps platforms
Every change goes through Git, every pull request gets its own copy of the app, and the pipeline catches mistakes before production does. It is the setup I run at home for 12 apps.
02
Django & Python web apps
Django apps with real tests, migrations checked in CI, and privacy and multi-tenancy designed in from the first model.
03
Data pipelines & scraping
Scrapers that assume the source will change. Scheduled jobs, validation, a saved snapshot of every failure, and clean data in Postgres or your warehouse.
04
AI & LLM integration
LLMs where they actually help: enrichment and extraction with a confidence score on every answer, agents that hand a human a pull request to review, and costs you can predict.
Something to build, or something broken
A few lines about what it is and where it runs is plenty.
Get in touch