I design and build the knowledge layers that make scientific data useful to AI.
That means OWL reasoning models grounded in upper ontologies like BFO and OBI, SHACL constraint layers as data contracts, knowledge graphs that connect instruments to decisions, and the FAIR data architecture that ties it together.
Most recently this was production work at AstraZeneca — an ontology for solid-state pharmaceutical chemistry, aligned to Allotrope, IDMP, and ChEBI, designed as the semantic backbone of a lab digital twin programme.
I've been doing this kind of work — under various job titles — for 25 years, across pharma, health, media and education. My USP is translating complex, ambiguous domains into clean, reusable, logically-rigorous data models that downstream systems can actually rely on.
Ontology & schemas — OWL 2, SHACL, LinkML, SKOS, RDF/Turtle
Upper ontologies & standards — BFO, OBI, OAE, IAO, CCO, QUDT, ChEBI, IDMP,
Allotrope, SNOMED CT, LOINC, FHIR
Knowledge graphs — Neo4j, TypeDB, TerminusDB
Lifecycle & tooling — ROBOT, pySHACL, GitHub-native schema workflows
Architecture — FAIR data products, data mesh, ontology-based data management,
agentic AI grounding
Recent clients: AstraZeneca, NICE, the Royal Pharmaceutical Society, and the UK Atomic Energy Authority. Earlier, Director of Content Architecture at Pearson International.
Human Sciences, Oxford (BA, First Class).