Stellenbeschreibung
Duration: 6 months+ | Start: 11/2026 | Remote: 100%
Beschreibung
IT
Data Analysis
Data Architecture
Neo4j
Azure Machine Learning
Semantic Web
SPARQL
Knowledge Representation
GraphQL
Artificial Intelligence Markup Language (AIML)
Description
Location: Remote (100%)
We are seeking a Senior IT Architect / Analyst to research, design, and evaluate methods for building high-quality knowledge graphs (KGs) and using them in AI applications — especially LLM and agent-based systems. You will work at the intersection of semantic modeling, information extraction, graph analytics, and AI evaluation: defining what "good" graph quality means, benchmarking the methods that produce graphs, and designing reasoning workflows that stay grounded in graph evidence.
This is a research-oriented architecture role: you will not only design systems but also establish how their quality is measured, proven, and reproduced.
Key responsibilities
Knowledge graph quality & measurement
Research, design, and evaluate methods for building high-quality knowledge graphs and applying them in AI systems.
Define and operationalize graph quality metrics: coverage, consistency, accuracy, provenance, and uncertainty.
Develop methods to measure completeness of graphs and detect information gaps.
Information extraction & benchmarking
Benchmark NER, entity resolution, relation extraction, and LLM-based extraction approaches for turning unstructured data into reliable graphs.
Evaluate methods for ontology development, domain modeling, and concept hierarchies.
Build benchmarks, baselines, and ablation studies; translate findings into reproducible experiments, evaluation frameworks, and practical recommendations.
Graph analytics & reasoning
Apply graph theory, statistics, and probabilistic methods to analyze structure, connectivity, communities, and information gaps.
Investigate traversal and retrieval strategies (hop limits, edge direction, weights, relevance scoring, pruning).
Design multi-step reasoning, critique, and verification workflows grounded in graph evidence.
Research link prediction and graph embeddings, keeping inferred knowledge explicitly separate from verified facts.
Required qualifications
Mathematical foundations
Strong grounding in graph theory, linear algebra, probability, statistics, and optimization.
Research depth
Demonstrated research experience in knowledge graphs, information extraction, ontology learning, semantic modeling, and graph-based machine learning (publications, dissertations, or equivalent applied research a plus).
AI-native experimentation
Practical ability to use LLMs and agent workflows to accelerate research, with a rigorous habit of validating results through evidence and measurement.
Data analysis & programming
Advanced Python; SQL; graph query languages (e.g., SPARQL, Cypher); reproducible research tooling.
Evaluation judgment
Rigorous experimental design: error analysis, uncertainty estimation, sensitivity analysis, ablation studies, and reproducible benchmarking.
Nice to have
Experience with enterprise knowledge graphs, RDF/OWL/SHACL/SKOS, and graph databases (Neo4j, ArangoDB, Virtuoso, GraphDB).
Experience evaluating or building GraphRAG / neuro-symbolic systems that ground LLM output in structured knowledge.
Familiarity with regulated or enterprise environments and provenance/audit requirements.
PhD or research-track background in semantic web, NLP, or graph ML.