Custom Software for Complex Domains
Thanks to AI coding agents, building custom software is faster and cheaper than ever, making it affordable even for smaller organizations. But this software must not only hold up in daily use, it also needs to fit how you actually work. That still requires solid software engineering and the drive to deeply understand your domain. I bring both.
The Challenge
- Off-the-shelf tools don't fit how you actually work, so your team juggles spreadsheets, copy-paste, and manual workarounds.
- Software built by an external agency is a black box, leaving you dependent on them for every small change.
- A vibe-coded prototype looked great in the demo, but breaks with real data and more users, and nobody knows how to fix one thing without breaking another.
- It's unclear where software or automation would actually pay off, so either nothing happens or money goes into the wrong project.
- Requirements stay vague because the people who know the domain and the people building the software don't speak the same language.
- AI gets added because it's exciting, not because it solves a real problem, resulting in unreliable features and surprising costs.
My Approach
Deep domain understanding plus solid software engineering, accelerated by AI coding agents.
The following principles shape how I work and are the reason the software solutions I build tend to actually get used, last, and leave teams stronger.
- Understand the problem before designing the solution. Software can only serve its users well if you understand what they actually need, so I spend a lot of time sitting down with domain experts and digging into the messy details before any solution is sketched. I'll also push back on anything that doesn't serve that need, which often includes AI features: sometimes a simple rule or a better interface beats a probabilistic model.
- Build fast without cutting corners. AI coding agents let me build in weeks what used to take months, but speed is worthless if the result falls apart. That's why quality is part of the initial design, not an afterthought: a clean architecture with security built in, well-designed data models, and a strong safety net of automated tests that catch issues early. I keep things simple and our options open instead of over-engineering, and as the software evolves with user feedback, I refactor continuously before workarounds pile up.
- Ship early and often. I work in small iterations, each ending with working software you can try out, so you can give feedback early and we can change course before going too far in the wrong direction. Web apps are deployed on managed platforms like Render, so you won't need a DevOps team to keep them running.
- Empower your team. You own the code and everything around it, including thorough documentation. I'm passionate about mentoring and love teaching teams how to make smaller changes themselves (with AI agents) and how to recognize when a change needs expert help. The goal is to leave you with everything you need to succeed after our engagement ends, though of course I'll remain available if you need me.
What I Offer
Whether you already know what you need or are still figuring out where software could help, I'm happy to support you.
Core Offering: Software Projects from Idea to Production
I take your project from the first conversation to running software: understanding your processes and requirements, designing the solution, building it end to end (frontend, backend, database, and AI features where useful), and deploying it on managed servers (optionally hosted in the EU). We work iteratively and in close exchange, so you see working software early and can steer as we learn. At the end, I make sure you're able to maintain and extend the software yourself.
- internal tools that replace spreadsheet chaos, e.g., for managing projects, applications, inventory, or donations;
- web apps and portals for your customers, members, or partners;
- automating repetitive workflows like data entry, document processing, or reporting, using large language models (LLMs, the tech behind ChatGPT) where it makes sense;
- dashboards that turn your data into a clear story and support actual decisions;
- turning a vibe-coded prototype into a robust product.
Entry Packages
- Process & Automation Check
For organizations that suspect their work could run smoother, but aren't sure where to start. Together with the people doing the work, we map out your key processes and identify where software or automation (with or without AI) would pay off, and where it wouldn't. You receive a prioritized list of opportunities with rough effort and expected benefit, so you can decide what's worth building. - Vibe-Coded App Rescue
For organizations whose AI-generated prototype doesn't hold up in practice. I review your app's architecture, data model, security, and usability, and give you an honest assessment of whether it's worth hardening or better rebuilt. You receive a prioritized roadmap, and if you want, I'll then make it production-ready.
Why Me
- 13+ years of experience, getting to grips with complex domains from chemical production (BASF) and cement manufacturing (alcemy) to industrial automation (TRUMPF & ALPLA) and highly regulated logistics (eFTI EU project with Toll Collect). I quickly learn how your work actually works and translate it into clear requirements.
- End-to-end software engineering skills: for example, I built Open Return, an AI-powered trip planner, from scratch to production, including the web frontend, a backend with an LLM pipeline, the database, and deployment.
- Deep machine learning and data expertise (PhD, published research, open-source libraries, years as a data scientist), so when AI or data features do make sense, they're built on solid ground.
- Experience explaining technical topics to non-technical audiences (50+ courses and workshops taught), so you'll always understand what's being built and why.
Ready to Work Together?
Whether you have an idea for a new tool, a prototype that needs to grow up, or just a hunch that your processes could run smoother — let's talk. I especially enjoy working with NGOs, startups, research labs, and small companies that are curious about new technology and keen to try things out.
Send me an email at [email protected] and we'll figure out how I can help.