Principal Business Lead - Enterprise Ontology and Semantic Layer | Irving, TX (Hybrid - 3 Days Onsite) | 6 Months Contract

Job Details:
Job Title:    Principal Business Lead - Enterprise Ontology and Semantic Layer
Location:    Irving, TX (Hybrid - 3 Days Onsite)
Duration:    6 Months Contract

Principal Business Lead - Enterprise Ontology and Semantic Layer
Operating Data Platform (ODP) - AI Semantic Layer & Agentic Automation

Description:
The client is standing up the Operating Data Platform (ODP), a new enterprise data lake consolidating operating, finance, supply chain, and commercial data from across the group's ecosystem businesses. ODP is the foundation for enterprise reporting, analytics, and AI automation. 
We are seeking a senior business leader to design, drive the building and own the enterprise semantic and ontology layer for the ODP: the governed definitions of the group's business entities and metrics, and the formal relationships between them, defined once and consumed by business intelligence, natural language query, and AI agents operating over the platform. The purpose of the layer is to make AI automation over ODP sufficiently reliable to act on, and the role is accountable for that outcome.
Basis for the role. AI success over enterprise data are predominantly semantic in addition to model success: right table, join, grain, and filter. An ungoverned agent returns a confident incorrect result at machine speed, upstream of human review. The semantic layer is the control that constrains that outcome.
The mandate is both hands-on and business-facing. The role works with the business to build the models, mappings, and agent-facing interfaces directly, and carries accountability for the semantic layer roadmap, the stewardship operating model, and adoption with the business.
•    Reporting line. Reports to the Chief Financial Officer.
•    Partners. Data Governance workstream, ODP platform engineering, the data security and access workstream, finance and FP&A, and the operating business data owners.

Responsibilities
The role carries four sets of responsibility - model the business, work with our partners to build the layer, govern it, and serve it to its consumers — together with leadership accountability across all four. The items below are indicative rather than exhaustive.
Model the business
•    Canonical business objects. Define the group's enterprise business entities — customer, vendor, product, contract, cost center, project, legal entity, asset — independent of any source system and maintain that independence against pressure to model the source rather than the business.
•    Conceptual, logical, and canonical modeling. Lead the modeling work and align it with the physical models owned by platform engineering, including conformed dimensions, declared grain, and permitted join paths.
•    Certified metrics. Define the certified metric set, including metrics reaching management and board reporting, with a single expression per metric and an explicit statement of what each excludes.
•    Taxonomy and ontology design. Author the taxonomies, controlled vocabularies, and formal ontology — classes, properties, relationships, constraints, and axioms — with a defensible position on where formal rigor is warranted and where a simpler model is sufficient.
•    Competency-question validation. Validate every model against the specific business questions it must answer, authored by the business rather than by the modeler.
•    Elicit semantics from the business. Extract definitions and business rules from subject matter experts across the operating companies and formalize them, including cases where functions have used the same term for different concepts.
Build the layer
•    Source-to-semantic mapping. Map ODP physical schemas into the semantic model and maintain the mappings as the platform evolves, using declarative mapping and transformation rather than bespoke code.
•    Entity and identity resolution. Design entity resolution across source systems and measure and report match precision rather than asserting it. Low-confidence matches must be visible as low-confidence to downstream consumers, including agents.
•    Knowledge graph construction where warranted. Build and populate the graph, with integration testing and automated validation in the pipeline.
•    Semantic model implementation. Implement the certified metric and entity layer in the platform's native semantic tooling and in the business intelligence layer, treating aggregate awareness and query performance as first-order requirements. Consumers route around a governed path that is slower than the alternatives.
•    Security inheritance. Ensure row-level security, masking, and classification are inherited and enforced through the semantic layer in concert with the data security and access workstream.
Govern it
•    Asset lifecycle. Establish creation, review, promotion, versioning, deprecation, and retirement for semantic and ontology assets, under version control with pull-request-based change review. Maintenance is a continuing obligation as the underlying business changes and is in scope alongside the build.
•    Change control and error containment. An incorrect definition in the semantic layer propagates to every consumer simultaneously. Own the change control, impact assessment, and regression testing that contain that risk, and the incident process for definition errors reaching production.
•    Validation beyond schema. Monitor the signal as well as the structure. A metric that continues to compile while an upstream default value has turned a filter into a no-op is a misstatement risk inherited identically by dashboards and agents. Distributional and semantic validation are in scope.
•    Federated stewardship operating model. Define object ownership, domain stewardship, promotion rights, and the arbitration path, balancing an operating company's specific need against enterprise reuse. This operating model determines whether the layer is sustained.
•    Catalog traceability. Maintain bidirectional traceability between the semantic layer and the enterprise data catalog and business glossary, so that a definition, its lineage, and its owner are discoverable from either direction.
Serve it to its consumers
•    Business intelligence. Lead the in the delivery of the certified semantic models consumed by the enterprise reporting layer, to a standard supporting SOX and enterprise data requirements where the metrics are financially significant.
•    AI and agent consumption. Expose governed entities and metrics to AI and agentic tooling through documented interfaces, including Model Context Protocol servers and query APIs, and define which tools and semantics an agent is permitted to call. Semantics are to be certified before agent access is granted.
•    Retrieval and grounding. Design ontology-informed retrieval and grounding for natural language query and retrieval-augmented generation and advise the AI and analytics teams on ontology-informed feature engineering and embedding strategy. The semantic layer and the vector store are complements.
•    Accuracy measurement. Establish and publish an answer-accuracy benchmark for natural language and agent consumption against a held-out, business-authored question set on the group's own schema, and report results on a fixed cycle.
Leadership accountability
•    Roadmap and sequencing. Own the semantic layer roadmap with business-facing accountability. Begin with the definitions carrying the most decision weight and scale by domain rather than pursuing a complete architecture before delivering value. Define current state, target state, and the transition architecture between them.
•    Team. Lead the team that is building and leading the semantic and ontology capability — semantic and ontology engineers, knowledge engineers, and semantic analysts — set the quality standard and develop the capability so the layer is not dependent on a single individual.
•    Technology selection. Evaluate and recommend semantic layer tooling, graph platforms, and query interfaces with a documented build-versus-buy position, a cost envelope, and a lock-in assessment. Manage selected vendors to scope and acceptance criteria.
•    Executive advisory. Serve as the domain expert to senior leadership on what the layer supports and what it does not and bring decisions forward with a recommendation and a quantified option set.
•    Measured outcome. Accountable for adoption, measured as the proportion of management- and board-reported metrics served from the layer, the count of surviving parallel definitions, definition reuse across data products, answer accuracy against the benchmark, and reduction in time spent reconciling numbers. 

Requirements
•    15–20 years in data architecture, data engineering, or enterprise information management, including 5 or more years with demonstrable ownership of an enterprise semantic layer, ontology, or knowledge graph in production. The band reflects enterprise architecture and leadership scope; the screening criteria are the depth and track record below rather than total tenure.
•    At least one comparable layer taken from build to adoption. Candidates must be able to describe a semantic or ontology layer they built, which consumers used it, what proportion of reported metrics it served, and what happened to the competing definitions it replaced.
•    Semantic and metric layer implementation. Hands-on delivery with enterprise semantic and metric tooling — Microsoft Fabric and Power BI semantic models, dbt Semantic Layer and MetricFlow, Cube, AtScale, LookML, Snowflake semantic views, or Databricks metric views — with depth in at least one and the architectural judgment to reason across the others.
•    Formal semantics. Working command of RDF, RDFS, OWL, SPARQL, SHACL, and SKOS, and experience with at least one production graph platform such as Neo4j, Amazon Neptune, Stardog, or GraphDB.
•    Dimensional modeling depth. Conformed dimensions, declared grain, slowly changing dimensions, and join path governance, demonstrated through design decisions.
•    AI and agent grounding. Direct experience making governed semantics consumable by large language models and agents — natural language query, text-to-SQL, retrieval-augmented generation, graph-based retrieval, and tool and context definition — with a clear view of what grounding does and does not correct.
•    Engineering credibility. SQL, Python, and modern software delivery practice: version control, code review, continuous integration, and automated testing applied to semantic assets.
•    Leadership and stakeholder scope. Track record leading a technical team while holding definitional authority across business functions, through governance and influence rather than line authority.
•    Governance and control literacy. Understanding of how certified metrics interact with financial reporting control, segregation of duties, and audit evidence where the layer serves financially significant numbers.
•    Executive communication. Written and verbal communication to a CFO-level standard, including explaining the business consequence of a semantic design choice in non-technical terms.
•    Education. Bachelor's degree in computer science, information science, mathematics, or a related discipline, or equivalent practical experience. An advanced degree in computer science, information science, artificial intelligence, data science, or formal logic is preferred and substitutable for additional years of relevant experience.

Preferred
•    Experience on a Microsoft data estate — Microsoft Fabric, Power BI, Microsoft Purview — and with semantic models sourced from Dynamics 365 Finance & Operations.
•    Reuse of an established industry reference ontology, for example GS1 for product and supply chain, or FIBO in a financial context.
•    Familiarity with the open semantic interchange specification now incubating at the Apache Software Foundation, and a view on portability of definitions between platforms.
•    Experience with an enterprise data catalog such as Collibra, Alation, or Microsoft Purview, and with Palantir Foundry object and ontology modeling, which is a related but distinct discipline from formal semantic web modeling.
•    Exposure to vector databases and embedding strategy as a complement to governed semantics.
•    CDMP (DAMA Certified Data Management Professional), Certified Knowledge Manager, or formal training in ontology engineering or semantic web technologies. 
•    Experience in private investment, technology infrastructure, or a comparable multi-entity, multi-jurisdiction group environment where the same business concept is defined differently in each operating company.