Senior Knowledge Graph Engineer

EcoVadis

7d•Hybrid

About The Position

Work smart, have fun and make an impact! EcoVadis is the leading provider of business sustainability ratings. Our solutions are backed by an international team of experts and powerful technology. We analyze data and build sustainability scorecards that give companies actionable insights into their environmental, social and ethical risks. Why apply to EcoVadis? Be a part of the global sustainability change in business. Grow your career. Work with extraordinary people. Feel valued for your contribution. Learn more about our team and culture on EcoVadis careers page. We are looking for a hands-on, production-focused Senior Knowledge Graph Engineer to join our growing AI Center of Excellence, responsible for using AI and machine learning to drive innovation across the organization. In this role, you will take the formal domain ontologies designed by our Knowledge Representation Architects and operationalize them into high-throughput, multi-hop systems. We welcome applications from European locations, with the possibility of remote work. Join us! Your work will directly power autonomous AI agents that solve complex sustainability challenges — including decarbonisation, sustainable procurement compliance, and supply chain resilience. You will bridge the gap between unstructured sustainability disclosures and structured graph databases, building entity-resolution pipelines that make enterprise data agent-ready.

Requirements

Degree in Computer Science, Mathematics, Engineering, or a related technical discipline
4+ years of production experience building and querying graph databases, specifically Labeled Property Graphs (Neo4j, Memgraph, TigerGraph) or RDF Triple Stores (GraphDB, Stardog, Virtuoso)
Strong experience in cloud technology, preferably Azure and its ecosystem (e.g., Azure Foundry, Azure Bicep, AzureML and Azure Cloud Storage)
Advanced proficiency in Python (RDFLib, NetworkX, PyGraphistry) for building scalable, production-grade data pipelines
Experience building entity extraction pipelines using modern NLP frameworks (LangChain, LlamaIndex, spaCy) or LLM-based structured extraction
Hands-on experience with modern data transformation tools (dbt) and integrating graph databases with vector stores (Qdrant, Pinecone, pgvector) for hybrid search architectures
Solid understanding of semantic web standards (RDF, RDFS and OWL, SKOS, SHACL, RDF-star, SPARQL), graph schema design principles (T-Box vs. A-Box separation), and mapping languages for dealing with heterogeneous data structures (RML, R2RML)

Nice To Haves

Experience working with domain-specific supply chain, carbon accounting (GHG Protocol), or lifecycle assessment (LCA) data structures is a plus
Direct experience building Model Context Protocol (MCP) servers to expose graph tools to LLM agents is a plus
Experience with enterprise OBDA approaches at-scale is a plus

Responsibilities

Graph Infrastructure and Ingestion Pipelines: Design, implement, and maintain high-speed GraphRAG ingestion pipelines that transform relational data (ERP, SQL), unstructured ESG reports, and streaming feeds into operational Labeled Property Graphs (Neo4j, Memgraph) and RDF Triple Stores.
A-Box Instantiation and Entity Resolution: Build automated Named Entity Recognition (NER), entity linking, and deduplication workflows to resolve mismatched vendor profiles, material SKUs, and facility coordinates into unified canonical graph nodes.
Semantic Federation and External Data Integration: Implement automated ETL/ELT pipelines to map and federate internal supply chain data with external, open-source ontologies and registries (such as GLEIF for corporate ownership, W3C SSN/SOSA for IoT sensors, and Copernicus for geo-hazard alerts, PROV-O for data provenance).
GraphRAG and Agent Tooling: Partner with AI/ML Engineers to build low-latency GraphRAG retrieval layers—writing optimized Cypher and SPARQL queries, implementing NL2Query tools for agents, hybrid vector-graph indexing pipelines, and Model Context Protocol (MCP) tool endpoints for autonomous LLM agents.
Deterministic Guardrails and Pipeline Validation: Operationalize SHACL (Shapes Constraint Language) shapes into automated data quality tests within CI/CD pipelines to prevent hallucinated or non-compliant data mutations from entering the enterprise knowledge graph.
Performance Optimization and GraphOps: Optimize multi-hop query performance, graph partitioning, and database indexing strategies to handle sub-second traversal over billions of nodes and edges.