Knowledge Graph and AI Engineer - #2169942

About the role

We are looking for an engineer to help build the knowledge graph and AI services behind a SaaS platform. The work involves turning documents and records into structured information, connecting related data and supporting search and matching with clear supporting evidence.

This role is open to both seniors and mid-level engineers with relevant practical experience. You will work with the AI lead, application developers and platform engineer, taking responsibility for agreed components as the product develops. Hands-on experience building knowledge graphs and working with ontologies is essential.

What you will work on

• You will develop RDF models and ontology modules for candidate and vacancy data, including skills, qualifications and employment history. The models need to preserve source evidence, record changes and represent conflicting information. You will write SPARQL queries, maintain taxonomy mappings and use SHACL to validate graph data.
• You will build Python services and repeatable processes to map structured records and extracted information into the graph. This includes incremental updates, duplicate handling and checks before data is published. Working with the wider team, you will ensure corrections and approved deletions reach the graph and search indexes, with tenant boundaries and access permissions respected.
• You will help develop information extraction services for documents, messages and transcripts, using FastAPI and schema validation. Extracted information should retain its source location and the model version used, with uncertain or unsupported results available for review. You will also contribute to entity resolution, using identifiers and similarity methods to find related records without losing the evidence behind a match.
• You will contribute to semantic and hybrid search, using embeddings, metadata filters and graph queries where appropriate. Search and matching results should be evaluated against representative examples, with explanations grounded in source evidence.
• You will write tests and maintain evaluation examples for the components you build. Alongside quality, you will consider latency, processing cost and failure handling. You will work with colleagues on service integration, investigate issues and document the decisions needed to maintain your work.

What you need

• We are interested in what you have built and can explain, rather than a fixed number of years in the role. You should have professional software or data engineering experience and be comfortable delivering tested components within a team.
• You must have personally implemented an RDF knowledge graph and designed or extended an ontology in a working system. This includes writing SPARQL queries, defining classes and relationships, modelling taxonomies and applying SHACL constraints. You should be able to explain your modelling choices and show how you handled data quality or query problems. Familiarity with graph terminology or querying an existing graph alone is not sufficient.
• Practical experience with at least one RDF graph store, such as GraphDB, Stardog, Apache Jena or Amazon Neptune, is required. You should understand stable identifiers, named graphs and provenance, and be comfortable using or learning relevant standards such as SKOS, PROV-O and JSON-LD.
• You need good Python and SQL skills, experience building APIs with FastAPI or a similar framework, and a working knowledge of schema validation and automated testing. You should be able to integrate services with application data and handle errors, retries and timeouts sensibly.
• You should have practical experience in some of the following areas: information extraction, entity resolution, embeddings, semantic search or retrieval evaluation. You do not need to have delivered every part of the AI stack, but you should be able to assess the quality and limitations of the methods you use.
• You will need to communicate technical decisions clearly and work closely with other engineers. An understanding of data privacy, access controls and the risks of sending sensitive information to model providers is expected. AI outputs must preserve uncertainty and follow the product’s agreed human review controls.

Useful additional experience

Experience with Azure, model evaluation tools such as MLflow, or search platforms such as Azure AI Search or OpenSearch would be useful. Other relevant experience includes document processing, ontology versioning, graph query optimisation, event-driven processing and deploying AI services. These are areas you can develop further in the role.

What we would like to hear about

Please describe a knowledge graph or ontology you have worked on, your personal contribution and the tools you used. We would like to understand a modelling or implementation problem you encountered, how you approached it and what you learned from the result.