Knowledge Graph Engineer

Babel Street is building an AI-native, agent-ready, API-first intelligence platform where computer vision, NLP, and document enrichment continuously produce signals about people, places, organizations, and events. The knowledge graph transforms those signals into a coherent, queryable representation of the world.

We are hiring a Knowledge Graph Engineer to help design, build, and operate the entity resolution and graph layer that connects our multimodal pipelines. You will work closely with the Director of Computer Vision and collaborate with ML, platform, and data engineering teams to connect mentions in text, objects in images, locations in geotagged media, and entities in structured sources into a unified graph.

This is a hands-on engineering role where you will contribute to production systems, grow your expertise in graph modeling and entity resolution, and help shape a platform that is still evolving. This will be a hybrid role based out of either our Reston, VA or Somerville, MA office.

• →Implement and evolve graph schemas and entity models for people, locations, organizations, images, and events
• →Build and maintain entity resolution pipelines that reconcile entities across NLP outputs (GLiNER, spaCy), computer vision outputs (object detection, OCR, face recognition, geolocation), and structured data sources
• →Develop ingestion pipelines that write resolved entities into graph systems such as ArangoDB and Spanner Graph or BigQuery Graph
• →Contribute to cross-modal linkage between text, images, and geospatial data, including confidence scoring and provenance tracking
• →Collaborate with computer vision engineers to integrate visual entity outputs into the graph layer
• →Help build graph-backed APIs and services used by downstream products and workflows
• →Write tests, contribute to design discussions, and support improvements to data quality and pipeline reliability

The knowledge graph is where Babel Street’s signals become usable intelligence. A detected landmark, a name extracted from text, an organization mentioned online, and a face identified in an image all become more valuable when they resolve to a shared representation of an entity.

In this role, you will help:

• Connect multimodal signals into a unified graph
• Improve the accuracy and consistency of entity resolution
• Enable downstream search, analytics, and agent workflows

• 2–5 years of professional software engineering experience (or equivalent experience through internships and projects)
• Proficiency in Python and at least one typed language (Java, Go, C#, or TypeScript)
• Familiarity with at least one graph database or graph processing framework (ArangoDB, Neo4j, TigerGraph, JanusGraph, NetworkX, or similar)
• Solid understanding of SQL and core data modeling concepts (keys, normalization, joins, indexing)
• Ability to work across teams, incorporate feedback, and make steady progress in ambiguous problem spaces

• Exposure to entity resolution, record linkage, or deduplication
• Familiarity with semantic modeling or schema design (RDF, OWL, property graphs, JSON-LD)
• Experience with cloud platforms such as GCP (Spanner, BigQuery, Cloud Run)
• Interest in computer vision, NLP, geospatial data, or multimodal systems
• Experience testing data pipelines, implementing data quality checks, or working with observability tools
• Exposure to search, ranking, or retrieval-based systems

• Month 1–2: Ramp up on the graph stack, contribute to existing pipelines, and ship initial changes
• Month 3–4: Take ownership of a component or pipeline with guidance, including testing and documentation
• Month 5–6: Deliver improvements to entity resolution quality, graph ingestion, or cross-modal linkage, and contribute to design discussions