Many promising ideas never make it into a product because nobody can say early enough whether they will actually work. A focused proof of concept answers that question with evidence instead of assumptions, before larger budgets and teams are committed.

I help companies turn new ideas into working prototypes and sound technical decisions. My background combines an international master’s degree in Vision and Robotics with years of industrial work in which research results had to survive real-world constraints, such as limited compute, real-time requirements and automotive safety standards. One of these results, an online camera calibration method, was granted as a patent in Europe, the US, Japan and China and went into series production.

What I can help with

  • Problem framing: translating a business idea or customer need into a clear technical question, measurable success criteria and a realistic scope.
  • State-of-the-art research: reviewing scientific literature, open-source projects and patents to identify the most promising approaches and the risks behind them.
  • Computer vision and perception: algorithms for camera calibration, motion estimation, object detection and scene understanding using OpenCV, ROS and TensorFlow.
  • Robotics and automated driving: behaviour estimation, trajectory planning and sensor-based algorithms, from simulation to integration on target hardware.
  • Rapid prototyping: building working prototypes in Python or C++, on desktop, embedded boards such as Raspberry Pi and Arduino, or in the cloud, accelerated with AI-assisted development tools.
  • Validation and benchmarking: designing evaluation datasets, metrics and validation tools that show objectively how well a solution performs.
  • Feasibility and trade-off studies: comparing approaches by accuracy, speed, memory footprint and hardware cost, so decisions fit both the software and the target platform.

How I work

  1. Understand the problem: clarifying goals, constraints, available data and what “success” means for your product.
  2. Research the options: evaluating existing methods and selecting the approach that best fits your requirements and hardware limitations.
  3. Build the proof of concept: implementing the chosen method quickly but cleanly, so it can evolve into production code if it succeeds.
  4. Evaluate and present: measuring results against the agreed criteria and presenting findings, limitations and recommended next steps.
  5. Iterate or hand over: refining the solution based on your feedback, or handing it over to your team with documentation and support.

Typical engagements

  • Evaluating whether a computer vision or AI idea is technically feasible before a product decision
  • Building a demonstrator for a customer pitch, trade fair or funding application
  • Comparing algorithm candidates for a new perception or calibration feature
  • Porting a research prototype towards a real-time, embedded-ready implementation

Why work with me

  • Research that ships: my calibration algorithm went from concept to a granted patent and series production in the Audi e-tron.
  • Academic and industrial background: an Erasmus Mundus master’s in Vision and Robotics, followed by years of automotive software development.
  • Production mindset: prototypes are designed with performance, testability and safety standards in mind from day one.
  • Clear communication: results are presented in a way that both engineers and decision makers can act on.

Let’s talk

Have an idea you want to validate or a technical question that needs a well-founded answer? Tell me about it and I will get back to you.

Get in touch