Projects

Here are some of the projects and themes I’m currently working on. My focus is on practical software solutions around automation, AI engineering, local LLM setups, and technological sovereignty in Europe.

Reminder Tool Link zu Überschrift

I’m currently working on a reminder tool for a client. The goal is to automate reminder delivery and track the responses in a reliable way, so follow-up processes do not need to be handled manually.

At its core, this is about building a workflow that makes operational communication more dependable: reminders need to be triggered at the right time, responses need to be captured clearly, and the current status needs to stay transparent. Because of that, the project is less about a single form or dashboard and more about a robust process solution.

Technically, the system is built with a React frontend and a TypeScript backend. I use Docker Compose for a reproducible local setup. The processing is designed around a worker architecture so that sending, response handling, and downstream jobs can run independently from each other.

What makes the project especially interesting to me is the combination of product logic and reliability. Automated reminders sound simple at first, but they become complex quickly once timing rules, response states, failure cases, and traceability come together.

Agent Memory Link zu Überschrift

Another topic I’m actively working on is agent memory. I’m testing different approaches to give AI agents not only short-term context, but also useful longer-term memory.

Right now, I’m especially focused on Hindsight. What interests me is how facts, experiences, preferences, and ongoing work context can be stored in a way that remains meaningfully retrievable and useful for agents later on. For me, the important part is not only storing information, but also the quality of retrieving it again when it actually matters.

Some of the questions I’m exploring are:

• What kinds of information should be stored permanently?
• How should memory be structured so it remains useful in real workflows?
• What role do retrieval, condensation, and knowledge updates play?

For me, this project is mainly a practical research field at the intersection of productivity, knowledge management, and AI system design.

Local LLMs for Security Research Link zu Überschrift

I’m also testing local LLMs for security research and as a foundation for a local coding agent setup. I’m interested in how far locally running models can go when privacy, infrastructure control, and reproducibility matter more than maximum convenience.

In my view, a local setup is especially interesting when working with more sensitive material, when experiments need to stay reproducible, or when different models and toolchains should be compared systematically. At the same time, it is a good way to understand where local models are already strong enough today and where clear limitations still remain.

Besides the security research angle, this topic also matters to me from a developer perspective. I want to build a setup that is genuinely useful for coding workflows without being fully dependent on external platforms.

European Software Solutions Link zu Überschrift

Another area I’m working on is European software solutions. I’m interested in which European alternatives already exist in important software categories and where meaningful gaps still remain.

The motivation is clear to me: stronger technological independence in Europe will become increasingly important over the coming years. That applies not only to infrastructure and cloud, but also to productivity software, AI tools, and digital platforms more broadly.

What interests me here is not only researching existing vendors, but also the perspective of how to help build better European alternatives myself. In the long run, I want to contribute to software that is technically competitive while also aligning better with the idea of digital sovereignty in Europe.