- Experience
- 5+ yrs
- Salary
- —
- Openings
- 1
- Posted
- 4 weeks ago
- Work mode
- In office
- Resume
- Required to apply
Where you'll work
Sign in to tell us what does and doesn't work for you here — it sharpens every match we show you.
Job description
Role Overview
This position involves building advanced natural language query capabilities over an Entity Graph using vector indexing and retrieval to support GraphRAG. The engineer will be responsible for ensuring responses are accurate, relevant, and properly evaluated.
Key Responsibilities
- Design and implement vector index strategies across graph data including chunking, embedding selection, and configuring indexes.
- Develop and maintain pipelines to generate, store, and update embeddings as graph data evolves.
- Implement GraphRAG by combining vector similarity search with graph traversal to enable context-enriched retrieval.
- Enable and refine natural language query scenarios, incorporating translation from natural language to graph queries.
- Establish evaluation frameworks to measure retrieval relevance and answer quality, and optimize retrieval settings accordingly.
- Ensure grounding of retrieved answers with source entity and edge traceability for provenance.
- Optimize performance related to index size, query latency, and compute/token cost.
- Maintain thorough documentation of retrieval architecture, evaluation results, limitations, and supported query patterns.
Required Qualifications
- Minimum 5 years in engineering roles with over 2 years of hands-on experience in vector search and retrieval-augmented generation (RAG) implementations.
- Proven track record delivering production-level retrieval pipelines (beyond prototypes).
- Strong proficiency in Python and practical experience with embedding models.
- Experience in assessing and enhancing retrieval quality through defined metrics.
- Knowledge of grounding techniques ensuring answers can be traced back to original data sources.
Preferred Skills
- Experience directly with GraphRAG combining graph and vector retrieval.
- Familiarity with Microsoft Fabric NL2GQL or Data Agent tools.
- Understanding of graph databases and traversal methodologies.
- Experience controlling large language model inference costs and latency at scale.
Primary Deliverables
- Vector index built over graph projections.
- Pipelines for embedding generation and updates.
- GraphRAG-based retrieval supporting defined natural language query use cases.
- Evaluation outputs and tuning documentation for retrieval performance.
- Support documentation covering query patterns and limitations.