Hey, colleagues! 👋 Welcome back to the “evening vibecoder” series, and today I brought you MyRepETL (GitHub link) — a tool for ETL through MySQL replication.

I’ll tell the story of how I solved the chaos in my Jupyter reports and built juport (Jupyter Report System) GitHub link. I’ll also share a few thoughts on how development is changing in the era of AI assistants.

I think no one will be surprised if I say that OpenAPI specs are awesome. It is common practice to write OAS spec first. But what if you already have a product?

Recently on one of my side projects the time came to start MVP to prove hypothesis. I needed a solution to fast start without my own infrastructure and decided to deploy the project to the cloud. Using kubernetes on a daily basis made me too lazy to use ordinary virtual machines so I glanced at managed kubernetes services.

One of my most long-playing projects is integration of the Camunda engine to corporate infrastructure. It is a promising technology to raise business processes to a new level. Most problems in the enterprise domain arise on information transition between employees. Some tasks could be done by one person and wait in the task list of another for a few days or even weeks before the second person will have time to work it out. Such inefficientness lengthens the process overall. If tasks could be done immediately things would run much faster.

Quick memo: How to turn on file swap on linux server. Today I worked with server where by some reason was turned off swap. It seems that this machine was targeted as k8s worker node but then was repurposed as database server. It is good idea to have swap turned on there.