Graph-Powered Job Matcher: From SQL Tables to Neo4j RAG in Production
summary
Shows a production pipeline that moves SQL HR tables into Neo4j, combines Cypher path reasoning with Neo4j’s native vector index and LangChain to build a graph-native RAG matcher that serves results in milliseconds and scales to hundreds of thousands of profiles without fine‑tuning models. Demonstrates ETL options (CSV LOAD, APOC+JDBC, Kafka Connect), hybrid retrieval (graph filters + embeddings) that boosts Hit@5 versus a fine‑tuned BERT baseline, encodes must-have vs nice-to-have in relationships, and costs pennies per thousand queries.
AI-written summary · about
chapters (9)
- 0:00Introduction to graph job matcher
- 1:15Why hiring still hurts
- 4:25Solving hiring challenges with graphs
- 7:22Moving SQL data to Neo4j
- 12:51Graph schema for job matching
- 15:46Retrieval Augmented Generation basics
- 17:55Graph RAG pipeline explained
- 20:06Cost and evaluation of the system
- 23:03Q&A and closing remarks
related talks
25mEnhancing Job Matching With Knowledge Graphs and RAG
13mOptimizing Real-Time Job Matching: Integrating Neo4j With RAG
30mPerformant GraphRAG Improving Neo4j to Drive Autonomous AI ...
01h 19mIntro to GraphRAG
40mRAG Pipeline Evaluation
39mEnhancing RAG with Neo4j Knowledge Graph
01h 17mGraph Database and Retrieval Augmented Generation
15mAgentic GraphRAG: Simplifying Retrieval Across Structured & Unstructured Data