Semantic Knowledge Architect in City of Westminster

The Semantic Knowledge Architect joins an established delivery team that includes an AI engineer responsible for agent development. The two roles work in close partnership: the AI engineer builds and maintains the agents; this role owns the knowledge layer those agents depend on.

Your responsibility is the quality, integrity and evolution of the semantic knowledge layer: the ontologies, taxonomies and knowledge graph structures that determine what agents know, how they connect information, and whether their outputs are accurate. You will expand and govern structures that are already in place, working iteratively as the programme develops and agent use cases grow. You will also be the diagnostic layer between agent outputs and the knowledge layer: when an agent produces an incorrect or incomplete output, you identify whether the root cause sits in the knowledge structure and fix it at source. This is a technically precise, high‑accountability role. The accuracy of agent outputs across the group depends directly on the quality of the knowledge layer you maintain.

The role follows a continuous iterative cycle across four activities:

• Expand: take new or evolving business and technical requirements and extend the existing ontology and taxonomy to accommodate them, maintaining consistency with the established model.
• Govern: manage the ontology and taxonomy as controlled, versioned assets with documented change rationale, ownership and review cycles.
• Validate: review agent outputs in collaboration with the AI engineer, identify gaps or inaccuracies that originate in the knowledge layer, and trace them to their structural source.
• Iterate: update ontologies, taxonomies and graph structures based on validation findings, closing the loop between agent performance and knowledge quality.

You will work closely with other team members for data context and domain knowledge. The data landscape is already understood within the team, so onboarding to the knowledge environment will be well supported.

What you will do:

• Extend the existing ontology to reflect new business requirements, additional entities and evolving operational concepts.
• Expand and maintain the enterprise taxonomy, ensuring classification remains accurate, consistent and fit for agent consumption.
• Own the governance framework for both the ontology and taxonomy: versioning, change control, documentation and review cadence.
• Work with the AI engineer to review agent outputs and identify where knowledge‑layer gaps or inconsistencies are driving errors.
• Update ontological and taxonomic structures in response to validated agent performance issues.
• Maintain the knowledge graph as an accurate, traversable semantic layer connecting group operational data.
• Ensure data ingestion into systems is governed by clear metadata and semantic standards.
• Document all structural decisions, changes and rationale to support long‑term knowledge asset governance.
• Contribute to the broader delivery team, sharing knowledge context with data, platform and business colleagues as needed.

Technical Pillars:

• Ontology Expansion & Maintenance: Build upon and extend the existing ontology foundation. This is not a greenfield task. The core domain model exists and the data is mapped. Your role is to deepen, refine and evolve it as operational requirements develop.
• Graph data modelling.
• Entity modelling and resolution.
• Semantic layer design.
• Graph databases: Neo4j, Stardog, GraphDB, Amazon Neptune.
• Data integration and linkage.

Key Deliverables – Ontology:

• Extended knowledge graph covering expanding operational domains.
• Semantic integration models connecting data sources accurately.
• Entity relationship frameworks that agents can reliably traverse.
• Graph integrity standards and validation processes.

Agent Output Validation & Knowledge Feedback:

• Ontology engineering and extension.
• Concept and semantic modelling.
• RDF/OWL/SKOS.
• Reasoning frameworks.
• Version control for ontology assets.
• Conflict resolution within existing models.

Key Deliverables – Agent Output Validation:

• Extended domain ontologies aligned to evolving business requirements.
• Refined concept models with documented change rationale.
• Versioned semantic schemas with change history.
• Updated enterprise vocabularies and definitions.

Taxonomy Governance & Development:

• Taxonomy design and iterative development.
• Governance framework design.
• Metadata modelling and stewardship.
• Faceted classification.
• Change management for knowledge assets.
• Knowledge organisation systems.

Key Deliverables – Taxonomy:

• Governed enterprise taxonomy with version control and change log.
• Taxonomy governance framework covering ownership, change process and review cycles.
• Metadata standards for data ingestion alignment.
• Documentation of classification decisions and rationale.

Knowledge Graph Integrity & Extension:

• Maintain and extend the knowledge graph as the operational semantic layer connecting data sources and AI agents. Ensure it remains accurate, consistent and fit for agent consumption as requirements evolve.
• RAG architecture understanding.
• GraphRAG.
• Semantic retrieval principles.
• Knowledge grounding.
• Agent output evaluation.
• Root cause analysis within knowledge structures.

Key Deliverables – Knowledge Graph:

• Regular structured review of agent outputs against expected knowledge.
• Documented root cause analysis for knowledge‑layer failures.
• Iterative ontology and taxonomy updates driven by agent performance.
• Shared feedback process with the AI engineer covering knowledge quality.

Essential Skills:

• Proven experience in ontology engineering, with demonstrated ability to extend and refine existing models rather than only build from scratch.
• Hands‑on capability with RDF, OWL or SKOS in a production or client‑facing context.
• Experience designing or governing enterprise taxonomies, including change management and version control.
• Ability to diagnose agent or system output issues and trace root cause to knowledge structure.
• Experience working collaboratively within a multi‑discipline delivery team.
• Strong documentation discipline: the ability to record decisions, rationale and change history clearly.

Highly Desirable:

• Experience with knowledge graph design and implementation using Neo4j, Stardog, GraphDB or Amazon Neptune.
• Familiarity with RAG architecture, GraphRAG or semantic retrieval as it relates to agent knowledge quality.
• Background in knowledge governance, metadata stewardship or information architecture.
• Exposure to financial services, private equity operations or similarly structured enterprise environments.
• Understanding of how AI agents consume ontologies and taxonomies and where structural gaps create output failures.

Scope and Boundary:

This engagement covers group‑level operations only. Fund management, investment decision‑making, portfolio company activity and fund‑level data are explicitly out of scope. The data environment is manageable in scale and already understood within the team. This is not a role that requires building a knowledge strategy from zero. It requires someone who can work precisely within an established framework, govern it rigorously, and extend it with discipline.

About Thebes Group:

Thebes Group is an optimisation company engaged on an AI transformation programme for a private equity group, focused exclusively on group‑level operations, not fund management or investment activity. We work with regulated industries and complex enterprises to reduce operational risk and build the foundations for intelligent transformation. This role offers the opportunity to do technically serious knowledge engineering work inside a live AI transformation programme, with a clear scope, a supportive team and direct accountability for the quality of what gets built. You will not be working in isolation. You will be working as part of a delivery team where your specialism is understood and valued.