Responsibilities
- Design and maintain LLM-powered extraction systems for parsing documents, identifying entities and relationships, resolving inconsistencies across segments, and detecting missing information using structured-output agents managed by durable workflows.
- Develop and evolve knowledge graph schemas modeled as typed Pydantic classes, define efficient Cypher query patterns, manage indexing strategies, and handle graph-level operations including schema migrations, while limiting scope to the graph layer itself.
- Build rule-based decision engines using table-driven logic where deterministic outcomes are preferred over probabilistic ones, ensuring clear interfaces between rule-based and AI-driven components.
- Implement data quality frameworks including schema validation, mandatory field enforcement, audit logging, and evaluation systems using expert reviews, automated checks, synthetic test data, and LLM-based assessment.
- Manage live data operations such as data backfills, synchronized schema migrations across relational and graph databases, monitoring of extraction performance and accuracy, and incident resolution.
Work Arrangement
Remote (Worldwide)
Other
- Fully remote, full-time position with flexible working hours.
- Availability during CET (Berlin time zone) is required.
- Potential for relocation following a successful long-term collaboration.
- Contract-based freelance engagement.
- Access to modern development tools, meaningful ownership of work, and a dedicated budget for job-related learning and training.
- Mission-driven work focused on enabling small and medium enterprises to comply with increasing security standards more efficiently.
- Application must be submitted via JOIN platform with a PDF CV and a brief (max 200 words) explanation of how you would architect a knowledge graph-supported RAG pipeline, covering ontology design, indexing, retrieval methods, and evaluation strategy.
