Hello and welcome to my personal website. I am a Solution Architect passionate about creating innovative solutions that solve real-world problems and enhance user experience. I try to create a strong connection between business and development teams so that the implementation effort is aimed at creating value for the customers. I make sure that the business initiatives I follow are carried out with the satisfaction of everyone, from the client to the team members.

Since 2023, I have been experimenting with solving the 227 Puzzle challenge hosted on the online judge platform using OpenAI’s GPT models. After countless trials, chatGPT cracked it when the o3-mini-high model was released in February. Why was this puzzle so challenging? The 227 Puzzle, a decades-old brain teaser, involves sliding tiles into place — deceptively simple yet complex to solve programmatically. The online judge platform hosts programming challenges that help users prepare for competitive programming contests.

Ever get tired of constantly adjusting your screen brightness depending on the lighting in your room? I did too. So, I had an idea: why not automate it? But instead of spending hours researching APIs and troubleshooting errors alone, I teamed up with ChatGPT to help me build a script that would do it all for me. What started as a simple question—“how can I change the screen brightness in windows 11 using python and winsdk?

One year later the experiment about creating a solution for the Architectural Kata Agile Dead Trees using chatGpt that you can find here I asked the same questions to GPT-4o and the results are quite impressive. At the end of the article I’m explicitly asking to create the software architecture taking into account microservices architecture design patterns, guess how it went :D This article was featured also on the Dear Architects Mailing list #191

This is an experiment to check the ability of ChatGPT (GPT-4) to create software artifacts from given requirements. The architectural kata is taken from Ted Neward’s repository tedneward/ArchKatas. The ChatGPT output was taken as it was generated without any modification, even the text formatting was generated in markdown language and left as it was generated. I will try to score the tasks from 0 to 10 considering how useful and precise is the generated output.