This article is not a technical deep dive into PostgreSQL, but rather a personal reflection on the culture shock I experienced when working with AI.
Recently, I was involved in a project to migrate a database from Oracle to PostgreSQL. Given my background, I assumed my responsibilities would be fairly standard:
- Assisting SQL developers in troubleshooting queries for PostgreSQL.
- Managing the data migration process from Oracle to PostgreSQL
However, on the very first day, the project manager dropped a bombshell on me: I was tasked with converting all Oracle queries to PostgreSQL by myself.
I flatly told her it was impossible. In my experience, a typical migration of about 5,000 queries usually requres a team of more than 10 developers working around the clock for over six months. There was no way a single person could handle this workload.
That was when the PM introduced me to Claude. She assured me I could pull it off using AI, specifically by feeding it the source code-what developers call the MyBatis mapper files. Since I had never managed or modified application mapper files before, the sheer scope of the task felt overwhelming, and I decided to leave the project.
Pressured by the PM's threats not to pay me, I was forced to take on the challenge. The PM helped set up my laptop with Microsoft Visual Studio Code, downloaded the mapper files, and gave me a crash course on how to commit and push the source files.
For a month, I practically lived inside Claude. The results were nothing short of mind-blowing. I was genuinely amazed at how accurately Claude understood my prompts and transformed complex SQL syntax.
Today, the developers are already testing the live application. Granted, there are still a handful of queries throwing errors that require rework, but look at the bigger picture: in previous paradigms, a massive team spent months converting these queries. Now, a single person using AI managed to convert an entire enterprise workload in just one month.
