1h ago · 9 min read · This post was originally published on basecut.dev. The standard GitHub Actions job for PostgreSQL tests looks something like this: - name: Set up test database run: psql "$TEST_DB_URL" -f scripts/s
Join discussion
5h ago · 5 min read · PostgreSQL Foreign Data Wrappers: Cross-Database Queries Explained Every few months I see a team build an elaborate ETL pipeline to sync data between two PostgreSQL instances -- scheduled jobs, temporary staging tables, conflict resolution logic, mon...
Join discussion
1d ago · 8 min read · Here we go again. This is my second post in a series documenting my transition into Backend + ML Systems Engineering. The first post covered building a fraud detection model from scratch. This one co
Join discussion
1d ago · 4 min read · PostgreSQL Logical Replication: Setup, Monitoring & Troubleshooting I've seen logical replication take down more primaries than it's protected replicas. Not because the feature is broken -- it works exactly as documented. The problem is that teams se...
Join discussion
1d ago · 12 min read · Introduction You spent three weeks building a RAG pipeline. The embeddings are clean. The retrieval scores look good. The demo works perfectly. Then it goes to production, and a week later the support team starts forwarding user complaints: the assis...
Join discussion
1d ago · 13 min read · The State Management Problem Most agent frameworks ship with a convenient default for state: a Python dict, a SQLite file, or an in-process object on the host machine. This works fine for a single agent running on a single machine during a demo. It s...
Join discussion
2d ago · 6 min read · The Problem My job board has a simple read path on paper: show “active” jobs, let users filter by location, remote/hybrid, specialty, and recency. In production it wasn’t simple. I had 8,000+ active listings, ~2,000 companies, and ~200 listing update...
Join discussion2d ago · 10 min read · Setting up a PostgreSQL staging database that actually reflects production is one of those tasks that sounds simple and turns into a day of cleanup. The obvious approach — pg_dump production to stagin
Join discussion3d ago · 6 min read · Every SMS verification reseller eventually hits the same wall: codes don't always arrive. In our production data over the last six months, the failure rate has been remarkably stable around 28-32% depending on country and service. Telegram in Indones...
Join discussion
