Graph Engineer

Salt Digital Recruitment; San Francisco (CA); On-site; USD 120,000 - 150,000; Full time

Job summary

Salt Digital Recruitment is seeking a Graph Engineer in San Francisco, California, to design and build knowledge graphs that enhance AI understanding of complex construction projects. Responsibilities include designing ontologies and converting documents into structured graph entities.

The ideal candidate will have strong experience in knowledge graph design, production experience with Neo4j, and the ability to optimize graph performance.

Qualifications

• Strong experience designing knowledge graphs in complex domains.
• Production experience with Neo4j or similar graph databases.
• Strong understanding of Cypher or Gremlin.

Responsibilities

• Design the core ontology and graph schema for construction knowledge.
• Build pipelines converting raw documents into structured graph entities.
• Optimize graph performance at scale.

Skills

Knowledge graph design

Experience with Neo4j

Graph query optimization

Python

Tools

Neo4j

AWS

Python

Job description

Graph Engineer

The Role We are looking for a Graph Engineer to design and build a knowledge graph that helps AI systems understand complex construction projects. This role is not about simply storing documents in a vector database. It is about modeling the relationships between tenders, specifications, drawings, RFQs, quotes, subcontractors, suppliers, materials, milestones, and regulatory requirements so AI agents can reason across project documentation in a structured and reliable way. You will work across graph databases, ontology design, document processing, vector retrieval, and AI systems to turn unstructured construction data into a queryable knowledge layer.

What You'll Do

• Design the core ontology and graph schema for construction knowledge, including entities, relationships, node types, edge semantics, and property structures.
• Build pipelines that convert raw documents such as PDFs, CAD files, spreadsheets, emails, specs, drawings, and quotes into structured graph entities.
• Extract entities, identify relationships, and resolve references across multiple documents.
• Design efficient graph traversal patterns for agent queries, including questions such as "What subcontractor quotes cover this specification?" "What penalties apply if this milestone is missed?" and "Which drawings, materials, and suppliers are connected to this requirement?".
• Combine vector-based retrieval with graph-based reasoning by using semantic search alongside relationship traversal.
• Optimize graph performance at scale, including indexing, caching, incremental updates, and large-volume relationship management.
• Support multilingual knowledge modeling across English, Dutch, and Japanese while maintaining semantic consistency.

What We're Looking For

• Strong experience designing knowledge graphs from first principles in a complex domain such as construction, engineering, legal, medical, financial, insurance, or supply chain.
• Production experience with Neo4j or similar graph databases such as Amazon Neptune, TigerGraph, or JanusGraph.
• Strong understanding of Cypher or Gremlin, graph query optimization, indexing, schema design, and memory management.
• Experience designing ontologies, taxonomies, or formal domain models.
• Strong understanding of RAG, vector embeddings, semantic search, and the limitations of vector-only retrieval.
• Strong Python skills and experience building scalable data pipelines for diverse document types.
• Comfort working in cloud environments, preferably AWS.

Preferred Experience

• Experience with construction, engineering, procurement, architecture, or other document-heavy technical domains.
• Background in NLP, named entity recognition, entity extraction, relationship extraction, or document intelligence.
• Experience using LLMs for structured extraction and reasoning.
• Experience with graph neural networks, knowledge graph embeddings, or hybrid graph/vector retrieval.
• Experience processing PDFs, CAD files, technical drawings, contracts, spreadsheets, emails, and regulatory documents.

Technical Environment

The current technical environment includes Neo4j, Cypher, Pinecone, ChromaDB, Python, unstructured.io, LlamaParse, custom OCR pipelines, Claude, GPT-4, Mistral, AWS, S3, RDS, ECS/EKS, PostgreSQL, Terraform, and CrewAI.

Ideal Profile

The ideal candidate is a hands‑on graph engineer who can design the knowledge architecture behind AI systems, build reliable data pipelines, and create graph structures that support complex reasoning across highly interconnected construction documents.

Salt is acting as an Employment Agency in relation to this vacancy.