Introduction: The Fever of Integration

We are currently living through a period of technological saturation. AI is no longer a specialized tool; it is an omnipresent force. From operating systems demanding background intelligence to job descriptions prioritizing prompt proficiency, the pressure to integrate AI into every corner of our workflow is relentless.

But in the rush to adopt the bleeding edge, we risk treating AI as an inherent solution rather than a powerful, albeit volatile, tool. For engineers, the current wave of “AI first” implementation carries serious, often overlooked risks that manifest not as immediate failures, but as creeping operational fragility and engineered debt.

The current setup for my blog offers great flexibility and control, source and CI/CD is on GitLab, DNS is handled by Cloudflare and server is VPS in Hetzner Cloud (2 vcpu/4 GB ram/40 GB disk) using Linux Fedora 44 on aarch64, but probably is overkill just to run Nginx and serve static content.

Migration

I’m migrating my blog to another platform with 2 main goals and one constraint, first goal is to reduce maintenance operations like apply OS updates, security patches, check logs, configurations updates, remove old logs, second goal is to reduce the costs to minimum, the constraint avoids any vendor lock-in or any major change on the blog tooling, currently using Hugo site generator.

Hexagonal Architecture (also known as Ports and Adapters) is one of the best patterns for keeping your core business logic pure, testable, and completely decoupled from external noise like databases, APIs, or UI frameworks.

To get started, I will explain these core concepts and how they are applied using FastAPI, PostgreSQL, and Pytest.

The core philosophy

The main idea is simple: Your business logic does not care what database or web framework you use. Instead of building your application in traditional database-centric layers, you put your Domain (Business Logic) at the absolute center. Anything outside that core (FastAPI, PostgreSQL, Celery, External APIs) is considered an “external detail.”

Sometimes I need a service to run when my user is logged in, otherwise the service can run like the other services when the machine start, systemd can provide that functionality.

Ollama

Ollama is one of the tools that allows you to run large language models locally. There are other tools like Unsloth Studio, llama.cpp, LM Studio, but Ollama is the simplest.

curl -fsSL https://ollama.com/install.sh | sh

I continue learning Go, so this time I created a password generator to generate secure passwords that exclude ambiguous characters

Features

• Generate multiple passwords at once.
• Customizable password length (minimum 8 characters).
• Excludes ambiguous characters (e.g., 0, 1, O, I, l, 5, S, q, g).
• Memory efficient

Implementation Details

To make the development easy and more maintainable creating a modern CLI tool I use spf13/cobra, Cobra framework and keep the dependencies at minimum with crypto/rand for cryptographically secure random number generation and strings.Builder for efficient string construction.

If you like the tiling window manager like i3wm and you want to switch from Gnome there are somethings missing like the network manager applet, the volume control, add/connect bluetooth devices and in some cases the gnome-keyring to handle passwords, private keys and certificates.

To fix that, there are some config tweaks and programs that would make the switch to i3 better. For this i going to use Fedora 34