Freelance AI & Software Design Consultant
Deep machine learning (ML) expertise plus software architecture skills, applied to complex technical domains. I help teams build AI and data-driven features that actually work in production and stay maintainable as the product grows.
The Challenge
Even experienced software engineering teams struggle with recurring issues when working on AI and data-driven features:
- AI features get bolted on as an afterthought instead of being a coherent part of the product, confusing users and creating architectural headaches.
- ML models fail to deliver because the problem and data aren't deeply understood, or the team lacks the theoretical grounding to anticipate common failure modes.
- Promising ML features stall as prototypes and never make it into production because MLOps practices like data monitoring and automated retraining pipelines are missing.
- Data visualizations fail to support decision-making because making complex information accessible takes specialized design skills most teams lack.
- Software products aren't designed for AI agents as users, lacking the agent-accessible interfaces, clean APIs, and validation layers agents need to interact with them reliably.
- Code and architecture get built around the technical solution — the ML model, the data pipeline — rather than the problem domain, making it hard to integrate functionality that addresses user needs holistically.
- New features take three times longer than they should because the codebase has accumulated too much technical debt and become a maze of workarounds and accidental complexity.
My Approach
A few principles shape how I work, and why the AI features I help build tend to actually ship, 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 make a point of sitting down with domain experts and digging into the messy details before any solution is sketched. For ML this is even more critical: unless you deeply understand how the data was generated and what it represents, the model won't work in the real world.
- Build AI features only when they add genuine value. Not every problem needs a probabilistic solution, and I'll push back when AI is overkill. Sometimes the right solution is a simple rule, a better interface, or no feature at all. AI should be used because it solves a real user need, not to give the marketing team another buzzword.
- Think production from day one. Designing for the future isn't actually that much more work, as long as you keep things simple and your options open instead of over-engineering. I always have deployment in mind from the start, especially for ML solutions, where the best prototype is wasted effort if you can't integrate the model and data pipelines into your product and workflows.
- Refactor continuously. As you work on a product, you learn things, and decisions that seemed sensible two months ago might now be obsolete. I push teams to refactor before workarounds pile up and the code becomes painful to touch. And when a codebase is already at that point, I help them chip away at the biggest problems with iterative refactorings rather than a risky rewrite from scratch.
- Empower the team. I'm passionate about mentoring, and want to leave your team with everything they need to succeed after our engagement ends.
What I Offer
I work with teams in a flexible, hands-on way — providing immediate value while setting them up for long-term success.
Core Offering: Embedded Consulting & Coaching
The main way I work is by joining your team part-time (typically 15–30h/week for 6–18 months, depending on the project). I support projects from design through implementation, shaping the architecture alongside your team, reviewing code, and refactoring where needed. The goal is always to leave your team stronger and more capable than I found them.
Entry Packages
To start working together, I offer two short, focused engagements (2-4 weeks) that provide immediate results and let us see whether a longer collaboration makes sense.
- AI Strategy Sprint
For teams starting a new ML or data-driven project — or evaluating whether AI is the right solution at all. We map out the problem space together, sharpen the product direction, and identify where AI and data-driven features genuinely add value (and where they don't). Optionally, we then sketch out a concrete technical approach — data requirements, model choices, architecture, and UX design — so your team is ready to start building. - AI System Audit & Roadmap
For teams with ML or data-driven features stuck as prototypes or causing pain in production. I review your codebase, data pipelines, ML workflows, and user-facing interfaces to identify weaknesses in architecture, code quality, maintainability, and usability. You receive a prioritized roadmap for refactoring and strengthening the system so it scales reliably into production and future features become easier to ship.
Why Me
- A rare combination of deep ML expertise (PhD, published research, open-source libraries) and software architecture skills — so AI features get built right, not just built.
- 12+ years of bridging complex domains and clean software, from process industry (BASF & alcemy) to industrial automation (TRUMPF & ALPLA), with hands-on insight into the challenges of applying ML in real-world industrial and R&D contexts.
- A track record of taking ML projects from messy prototypes to production systems that deliver real-world impact.
- Proven results in taming accidental complexity, simplifying codebases and getting teams back to shipping features without headaches.
- A strong focus on knowledge transfer through pairing, code reviews, and training (50+ courses taught), so your team is set up for success after I leave.
Ready to Work Together?
Whether your team is starting a new ML or data-driven project, struggling to move a prototype into production, wrestling with a codebase that's become hard to maintain, or trying to figure out how to get ready for a future with AI agents — let's talk about what you're working on.
Send me an email at [email protected] and we'll figure out how I can help.