Skip to content

README.md

Latest commit

 

History

History
63 lines (45 loc) · 2.43 KB

README.md

File metadata and controls

63 lines (45 loc) · 2.43 KB
Raw

graph_tutor

An AI-agent-tutored curriculum for learning graph and vector databases by building a CS/AI/ML knowledge base in Neo4j.

What You'll Learn

12 progressive lessons that take you from graph fundamentals to GraphRAG:

# Topic
1 Graph Databases 101 — nodes, edges, labels, the graph data model
2 Neo4j Setup + Your First Nodes — CREATE, properties, labels
3 Relationships — directed edges, relationship types, topology
4 Querying with Cypher — MATCH, WHERE, filtering, projection
5 Building the Knowledge Graph — MERGE, UNWIND, schema design
6 Pattern Matching and Traversal — variable-length paths, subgraphs
7 Graph Algorithms — centrality, PageRank, community detection
8 Vector Search Fundamentals — embeddings, semantic meaning, distance
9 Neo4j Vector Indexes — HNSW, index configuration
10 Similarity Search — KNN queries, semantic retrieval
11 Hybrid Graph + Vector Search — combining structure and semantics
12 GraphRAG — retrieval-augmented generation with graph context

How It Works

An AI tutor agent guides you through each lesson, explaining concepts before code, checking your understanding, and building a running knowledge base that grows across all 12 lessons. No skipping fundamentals — everything is explained from first principles.

Stack

  • Database: Neo4j 5.x+ (run via Docker)
  • Query Language: Cypher
  • Plugins: APOC, GDS
  • Interface: Neo4j Browser (localhost:7474) or cypher-shell

Prerequisites

  • Docker (for running Neo4j)
  • Neo4j 5.x+ with APOC plugin
  • No prior graph database experience needed

Quick Start

# Start Neo4j
docker compose up -d

# Open the browser
open http://localhost:7474

Then follow the lessons in TUTOR.md with the AI tutor agent.

Project Files

  • TUTOR.md — The 12-lesson curriculum and agent system prompt
  • AGENT.md — Cypher conventions, Neo4j setup, and teaching rules

Design Decisions

  • CS/AI/ML knowledge base as the running example — you're building a graph about the domain you're studying
  • Neo4j native vectors over external vector DB — one tool to learn, demonstrates graph+vector integration
  • Pre-computed embeddings over live API calls — focus stays on database concepts, not ML plumbing
  • GraphRAG as finale — shows the convergence of graph and vector search in modern AI applications