Head of Test Engineering at Buddy · Bielsko-Biała, PL
I'm the person who has to trust the pipeline before anyone else does. Our suite runs on every branch, and most of the job starts when it goes red: telling a real bug from a flaky test, fast enough that nobody starts ignoring the results.
At work
- Regression triage — a suite that runs on every branch, and a first pass that sorts whatever came back red into product bug / test fix / flake / infra. Small cheap models do that pass; the expensive ones only see what's left.
- Run analytics — where the time goes, which tests actually fail, and which ones are unstable over a longer window. Dashboards and reports instead of guessing from the last red build.
- Shorter feedback loops — running what a change actually touches instead of everything, and keeping results fast enough to arrive before people switch context.
- Agent tooling — the QA work that used to be manual, handed to agents: triage, log digging, test selection. Mostly Claude Code, MCP and a pile of small skills that encode how we actually work.
- QA tooling — making all of the above readable for people who don't write tests.
I test Buddy with Buddy, so most of what I write about comes from running our own CI/CD day to day rather than from a slide.
Off the clock
Local-first home automation — Home Assistant, ESPHome, a Proxmox box doing the heavy lifting, lights and blinds running on hardware I can still flash myself. I keep landing in the same place: if it needs someone else's cloud to turn on a light, it isn't finished. Same instinct as at work, really — I don't trust a result I haven't verified, whether it's a green test or a datasheet.
Also, Barça Culé.