Turn data into knowledge AI can reason with.
Knowledge engineering, ontologies, and graphs for healthcare AI solutions.
The gap
Most healthcare data describes what. It rarely captures what it means. A patient record, a lab result, a device output — each one stores facts, but the meaning lives in people's heads and in disconnected systems. That's the gap where AI projects stall: a model can pattern-match across rows, but it can't reason about a diagnosis, respect a clinical guideline, or explain why it reached a conclusion. Knowledge engineering closes that gap by giving your data an explicit layer of meaning — concepts, relationships, and rules that both machines and clinicians can work with
What I do
From "is this even possible?" to a working, explainable system.
I work with healthcare and life-sciences teams at whatever stage they're at. Some need a clear-eyed feasibility assessment before committing budget. Some have a green light and need someone to design and build the ontology or knowledge graph properly the first time. Others have the system and need their people trained to own it. I cover all three — feasibility, build, and enablement — and I'm equally comfortable talking to your clinical leads and your data engineers.
Why me
I've worked on both sides of the chart.
Before I moved into data, I worked inside hospitals — so I understand clinical workflows, terminology, and why trust and traceability aren't optional in this field. On the engineering side, I hold a PhD in biomechanics and biomedical engineering, a master's in complex systems engineering, and a kinesiology background that grounds all of it in human physiology. That combination means I don't just model healthcare data correctly; I model it the way the domain actually behaves — which is usually where generic data teams come unstuck.
Why it matters in healthcare
In healthcare, "the model said so" isn't good enough.
Clinicians, regulators, and patients all need to know why a system reached a conclusion, what evidence it used, and whether it can be trusted at the point of care. Knowledge graphs and ontologies make AI auditable: every inference can be traced back to a concept, a source, and a relationship. That's the difference between a black-box prediction and a decision you can stand behind — and increasingly, it's the difference between a pilot and a system that's allowed into production.
How to start
Not sure whether knowledge engineering fits your project? That's exactly where a feasibility conversation helps.
The cheapest mistake to avoid is building the wrong thing well. A short feasibility engagement tells you whether semantics and graphs will genuinely move your AI solution forward, what it would take, and where the quick wins are — before you commit serious budget. Get in touch to start talking through what you're working on.
Feasibility & Roadmapping
Before anything gets built, I assess whether semantics and knowledge graphs will actually improve your AI solution — and where they won't. You get an honest picture of value, effort, and risk, plus a roadmap you can take to stakeholders.
Ontology & Data Modeling
I design the conceptual backbone: the concepts, relationships, and constraints that capture how your domain really works. Whether it sits over a relational model, as an RDF graph or as a property graph, the model is built to be reused, not rebuilt.
Knowledge Graph Engineering
I build production ready knowledge graphs — in Neo4j, RDF triplestores, or hybrid setups — that connect siloed data into something AI can traverse and reason over. Relationships become first-class, queryable, and explainable.
Semantic interoperability
Healthcare runs on standards. I align your data with the terminologies and frameworks that make it interoperable — SNOMED CT, ICD, LOINC, FHIR, OMOP — so your knowledge layer speaks the same language as the wider ecosystem.
Explainable & Traceable AI
I ground AI in your knowledge graph so its outputs can be traced to sources and concepts — including graph-grounded retrieval (GraphRAG) that reduces hallucination and gives LLMs a factual backbone clinicians can audit.
Training & Enablement
A system nobody understands is a liability. I train your data and clinical teams to read, extend, and maintain the knowledge layer, so the capability stays in-house after I leave.
Still have questions?
Get in touch
Adresse : 200 rue de la Croix Nivert, 75015 Paris.