Get in touch
The problem, the data you have, and your timeline — that's enough to start.
Designing and shipping production-grade segmentation and object detection pipelines — from data preparation and model architecture to evaluation (facial wrinkles, skin lesions, medical imaging, drones, quality inspection in manufacturing).
Fine-tuning and deploying and evaluating vision models plus diffusion models (LoRA, Stable Diffusion, FLUX) for real product use cases. Practical guidance on when GenAI is the right tool, and when it isn't.
Over 20 years of coding experience (started coding at 15 years old without AI assistance) and hands-on leadership of technical teams delivering AI/ML solutions across industrial projects. Practical guidance on deploying and maintaining ML systems at scale (AWS Cloud).
Quantitative CT/MRI and ultrasound image analysis and image processing grounded in a PhD in Biomedical Engineering and clinical research across European hospitals, including work that contributed to a commercial medical device product.
Over 20 presentations delivered to audiences ranging from medical conferences to industry events, including AWS Summit 2024 and a special event at the University of Bern. Available for conference talks, technical workshops, and panel discussions on AI/ML.
Literature research and review, plus writing or reviewing scientific articles on request — grounded in a PhD and a track record of peer-reviewed publications in biomedical engineering and medical imaging.
Engagement models
Days to a few weeks
Feasibility reviews, model and architecture critiques, build-vs-buy calls, and honest answers on whether AI is the right tool for the problem at all.
Weeks to months
Embedded delivery: data pipeline, model, evaluation and deployment — taking a promising demo through to something that survives production conditions.
One-off or recurring
Workshops, mentoring, code and research reviews, and internal talks that leave your team able to keep going without me.
Most projects start with a short call, a written problem statement and a small scoped first step — so we both know whether the collaboration works before anyone commits to a long engagement.