Production database engineering

AI DBA is a publication about running databases that cannot afford to fail.

It covers PostgreSQL and ClickHouse in production, the change data capture pipelines between them, and the automation worth trusting around both. The focus is deliberately narrow: the parts that are hard to get right and expensive to get wrong — delivery guarantees, recovery paths, schema change, failover, and the signals that make silent data loss visible before a customer finds it.

What to expect

Articles are written from work on real systems rather than from documentation. Where a piece describes a pipeline, its configuration is published as a runnable reference repository, so you can read the thing itself instead of reconstructing it from prose.

They are long, specific, and revised as systems change. They favour the operational detail that usually gets cut: what to monitor, what fails first, and what the recovery procedure actually looks like when you need it.

What this is not

This is not vendor content and not a tooling scoreboard. It does not cover systems that have not been run in production, and it does not recommend an architecture without saying what it costs.