Director of Ontology Knowledge Engineering

Knowledge Graph Engineering, Financial Risk Data

Knowledge Graph | Semantic Modelling | Graph Databases | Credit Risk | AI

Head of Knowledge Graph Engineering - Data Science and Analytics

London (hybrid)

Up to £200k + B&B

This is a knowledge graph engineering, semantic modelling, graph query design, risk data, graph-grounded AI role. Knowledge graph engineers, data engineers, applied data scientists or semantic technologists with a knowledge graph or connected data application built for analytics, content or AI, taken from domain conversations to a working implementation. You will be tasked with owning what a knowledge graph is in a new data science and analytics organisation: semantic model, data mappings, queries, provenance and temporal rules, validation with domain experts, and the first working implementation.

THE ROLE:

• Graph build: initial graph models, data mappings, queries and working implementations, built personally from a live product problem and released to users


• Languages and tools: Python, SQL, Cypher, SPARQL, RDF, property graphs; no single stack required


• Semantic model: analytical domains, subjects, capabilities, methods, inputs, outputs and the relationships between them


• Relationship modelling: identity, time, provenance, validation; meaning, timing, source and limits of each relationship stated


• Risk pathways: country and industry developments linked to companies, exposures and credit outcomes


• Graph and AI: connected retrieval, grounded answers, traceable sources, answer quality assessed against real questions


• Model testing: graph relationships, features and algorithms tested with data scientists against a simpler baseline
  
• Content enrichment: tagging with research and content teams linking documents, topics, entities, events and analytical capabilities; source and context preserved


• Delivery: ingestion, validation and interfaces with engineering and platform teams; shared entity and identifier services; entity-linking requirements; reusable patterns; teaching partner teams

Own the first implementation: a working graph for one priority risk or analytics question, on real data, with a model validated by domain experts and prospective users. Own the test: at least one adjacent use case, such as graph-derived features for a model or connected research for source-backed AI, measured against a simpler approach. Own the patterns: modelling patterns, working examples and an adoption plan for other teams.