For engineers building retrieval systems that need relationships, not just similar passages: model and build knowledge graphs, extract them with LLMs, and use them in GraphRAG pipelines you can evaluate and operate.
The curriculum is being prepared and reviewed. Enrollment is not open yet.
About this course
Vector retrieval finds passages that resemble a question, but it struggles with questions that require following relationships across documents or summarizing a whole corpus. Knowledge graphs make entities and relationships explicit, and GraphRAG uses them to retrieve connected context. This course covers both the long-established discipline and the new LLM techniques. You start with data models: labeled property graphs and Cypher (including Neo4j's Cypher 25 and its alignment with the ISO/IEC 39075 GQL standard), and RDF, OWL, SPARQL and SHACL for semantic graphs, with Amazon Neptune, Neo4j, Memgraph and Graphwise GraphDB (formerly Ontotext GraphDB) as reference stores.
You then build graphs from unstructured text: schema-guided LLM extraction, entity resolution and canonicalization, provenance and incremental updates. The GraphRAG modules implement Microsoft GraphRAG's indexing pipeline with Leiden community detection and its local, global, DRIFT and basic search modes, compare lighter-weight approaches such as LazyGraphRAG and LightRAG, and build text-to-Cypher and hybrid graph-plus-vector retrievers with the Neo4j GraphRAG Python package, LlamaIndex PropertyGraphIndex and LangChain integrations. The course ends with evaluation, cost and operations: when GraphRAG is worth its indexing cost, how to measure it against a strong vector baseline, and how to keep a graph correct as sources change. Facts are stated as of October 2026. It ends in an automated exam; passing earns a Wirewalk Institute of AI certificate of completion and assessment.
What you will be able to do
Choose between a property graph and an RDF model and design a schema or ontology for a retrieval use case
Write Cypher, GQL-aligned and SPARQL queries for multi-hop retrieval
Build a knowledge graph from documents with schema-guided LLM extraction, entity resolution and provenance
Implement Microsoft GraphRAG indexing and explain when to use its local, global, DRIFT and basic search modes
Build text-to-Cypher and hybrid graph-plus-vector retrievers with safeguards against invalid or unsafe queries
Evaluate GraphRAG against a vector RAG baseline on accuracy, comprehensiveness, latency and indexing cost
Operate a graph-backed retrieval system with incremental updates, access control and quality monitoring
Syllabus
6 modules · 21 lessons · about 20 hours
1
Module 1: When Graphs Beat Vectors
The question types that break vector RAG and what a graph adds.3 lessons
Multi-Hop, Aggregation and Global Questions: Where Vector RAG Fails
Knowledge Graphs, Ontologies and Taxonomies: Terms and History
A Decision Framework: Vector RAG, GraphRAG or Both
2
Module 2: Graph Data Models and Query Languages
Property graphs and RDF, and how to query each.4 lessons
Labeled Property Graphs and Cypher 25 in Neo4j
GQL (ISO/IEC 39075): What the Standard Changes for Graph Queries
RDF, OWL, SPARQL and SHACL for Semantic Graphs
Choosing a Store: Neo4j, Amazon Neptune, Memgraph and Graphwise GraphDB
3
Module 3: Building Graphs from Unstructured Data
LLM-driven extraction that produces a graph you can trust.3 lessons
Schema-Guided Entity and Relation Extraction with LLMs
Entity Resolution, Canonicalization and Linking to Reference Data
Provenance, Confidence and Incremental Graph Updates
4
Module 4: GraphRAG Architectures
Community-based and lightweight graph retrieval methods.4 lessons
Microsoft GraphRAG Indexing: Entities, Leiden Communities and Community Reports
GraphRAG Query Modes: Local, Global, DRIFT and Basic Search
Lower-Cost Variants: LazyGraphRAG and LightRAG
Prompt Tuning and Cost Control for Graph Indexing
5
Module 5: Graph Retrievers in Practice
Querying graphs from LLM applications and agents.4 lessons
Text-to-Cypher and Text-to-SPARQL: Schema Prompts, Validation and Read-Only Execution
Hybrid Retrieval: Vector Entry Points with Graph Traversal
Neo4j GraphRAG Python, LlamaIndex PropertyGraphIndex and LangChain Graph Integrations
Graphs as Agent Tools and Memory
6
Module 6: Evaluating and Operating GraphRAG
Proving the graph earns its cost and keeping it correct.3 lessons
Evaluating GraphRAG Against a Strong Vector Baseline
Graph Quality Metrics: Coverage, Duplication and Constraint Violations
Access Control, Freshness and Scaling a Production Knowledge Graph
Exam and certificate
The exam unlocks once you have completed every lesson. It draws 25 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 60 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.
Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.
The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by Neo4j, Microsoft, Amazon Web Services, LlamaIndex, LangChain, Graphwise, or Memgraph. Product and company names are trademarks of their respective owners.
A hands-on course on building no-code assistants with the builders already in your workplace tools: ChatGPT projects, custom GPTs and their successor plugins, Claude Projects and Skills, Gemini Gems, and Microsoft 365 Copilot Agent Builder and Copilot Studio. You leave with a tested, documented assistant and a plan to share it safely.
A hands-on, code-first course on the three major LLM APIs side by side: messages, streaming, structured outputs, tool calling, vision, cost, caching, batch, retries, abstraction layers, cloud routes and testing. You finish by building and hardening a small multi-vendor service.
A practitioner's course on retrieval-augmented generation: when to use it, how to ingest, chunk, embed and index documents, how to combine BM25 and vector search with reranking, and how to measure, debug and operate the result. Every module includes working Python sketches.