AI Graph Engineering - LLM LangChain - Python, Neo4j & RAG
Licensed sponsor United Kingdom Contract Posted 1 hour ago
We are building an enterprise-grade AI platform to enable secure, scalable and production-ready use of Generative AI and ML across the Customs Declaration Service.
Location: Anywhere in the UK, with the option to attend the office when required.
Mandatory
- Building and populating knowledge graphs programmatically from structured and unstructured sources, not UI-based curation
- Graph databases/query, e.g. Neo4j/Cypher or RDF/SPARQL
- Python data engineering for ingestion/ETL at volume
- Vector embeddings and chunking strategy
Nice to Have
- LangGraph or equivalent orchestration experience
- AWS-native AI stack, Bedrock, OpenSearch; other cloud stacks considered if transferable
- Regulated/government environment experience
Key Responsibilities
The Knowledge Engineer builds and operates the pipelines that populate and maintain the CDS knowledge graph at volume, turning the Knowledge Modeller's schema and platform-team retrieval design into a working, governed data asset.
- Build ingestion pipelines to populate the knowledge graph programmatically from structured and unstructured sources
- Implement and maintain graph population workflows, entity/relationship extraction, loading and validation against the defined schema
- Own chunking and embedding strategy for content feeding retrieval
- Monitor and maintain data quality, freshness and lineage within the knowledge graph
- Support reconciliation processes to catch ingestion gaps, e.g. missed webhook events, without adding load on source systems
- Work with AI Engineers to ensure the graph and retrieval layer serve RAG/agent consumption correctly
- Contribute to evaluation of graph/retrieval quality from a data-completeness perspective, distinct from AI Engineer's model-output evaluation
