DEV Community

Cover image for Reflections on System Architecture & Infrastructure: What Really Matters
Schrödinger Hat
Schrödinger Hat

Posted on • Originally published at schroedinger-hat.org

Reflections on System Architecture & Infrastructure: What Really Matters

Hi there! I'm Pato and today I have another article for you today:

I have recently attended an event called "Software Architecture & Infrastructure" organized by Tacto where the CTO of DeepL Sebastian Enderlein and Tacto Co-Founder & CTO Nico Bentenrieder got me thinking about what makes systems and teams truly work. Without further ado let me share what I learned:

The Power of Observability

First off, let's talk about observability. It's not just another tech buzzword; it's really important to have solid processes to analyze what can go right or wrong in your system. Sebastian and Nico found that the more metrics you can compare, the more robust your understanding becomes. It's like having multiple camera angles on a sports play – each perspective adds to the complete picture.

Building Strong Core Systems

On the technical side, when talking about core systems especially in frontend engineering, you need to have a good foundation of components that teams can work with without too much hassle. Once that structure is there, you can expand vertically. But here's the thing – it totally depends on what kind of app you're building. If your product needs to be heavily customizable, you need to build flexibility into your DNA and be ready to adapt quickly.

The Truth About Hiring and Culture

Speaking of adaptation, let's talk about hiring. You know what's funny? The most important thing isn't the technical skills – it's the cultural fit. The person will show you almost immediately if they're a good fit for your company. What really matters is understanding their motivations, their purpose, their curiosity. Sebastian and Nico have seen it firsthand: people with a closed mindset like "I'm only backend" or "I'm only frontend" or "I'm only DevOps" – that just doesn't work anymore. You need people who are open-minded and ready to learn.

Silicon Valley vs. Europe: A Tale of Two Ecosystems

Now, here's something interesting about founding a business in different regions. Europe is really complicated, but once you understand the intricacies or manage to open your business, it's actually way better than you might think. The talent density is just different from Silicon Valley. In SV, you have way more engineers who are all about putting products forward – they basically give everything they've got for that. Europe is different – companies can't just search in their own country because the US market is way bigger in comparison. You have to adapt when searching for talent across countries and when expanding your business internationally.

The Art of Progress: Less Planning, More Adapting

Let me share something about progress that they have learned the hard way: don't overthink the processes. Yes, processes are necessary and they push the business forward, but the worst thing is when you try to overplan and overthink every single aspect. Then something unexpected happens – could be political, technical, or cultural – and your system can't adapt to it. You've basically sent your business to die. It's better to have a balance between efficiency and growth. You can't have too much of either for too long, or things start affecting your team's output.

Specialized vs. General AI: A Real-World Comparison

Here's something fascinating I learned about AI models. When you compare ChatGPT with models specifically trained for a purpose, like DeepL, the specialized ones are way more efficient. DeepL, for example, has this clever way of training glossaries and specific translations. Take the BMW slogan "Freude am Fahren" – if you give it the proper BMW context, it translates to "The Ultimate Driving Machine." Try that with a normal LLM, and you'll just get "Joy of driving." What makes this interesting is how DeepL lets you add glossaries about specific terms and remembers your preferred translations.

Final Thoughts

The key takeaway from all of this? Whether it's system architecture, team building, or AI implementation, success comes from finding the right balance between structure and flexibility. You need enough foundation to be stable, but enough adaptability to evolve. And most importantly, you need people who understand and embrace this balance.

Note: Credits of the image go to Tacto Technology GmbH, Sebastian Enderlein and Nico Bentenrieder.

Top comments (0)