Getting Started with AstraeaDB
AstraeaDB is a database that stores your data as a graph, which is a network of items and the connections between them, and can also find items by meaning rather than by exact keyword. It is written in Rust and runs on your own machine.
By the end of this path you will have built a knowledge graph from raw text, searched it by meaning, answered questions against it with a language model that can only cite what the graph actually contains, and used the same techniques to investigate fraud and intrusion.
Start at the beginning if graphs are new to you. If you have used a graph database before, skim Crawl and start at Walk.
Choose your language
Crawl runs as two parallel tracks. Pick one and the lesson lists below follow you through the site. You can change your mind on any page.
Crawl
Get a server running and learn the basics. Pick Python or R and stay in it.
After this tier: You can start AstraeaDB, create nodes and edges, run a vector search, and see your graph in a browser.
- Why Graphs, and Why Now: An Introduction to AstraeaDBboth
- Getting Started with AstraeaDB in Pythonpy
- Getting Started with AstraeaDB in Rr
- Vector and Hybrid Search with AstraeaDB in Pythonpy
- Vector and Hybrid Search with AstraeaDB in Rr
- Graph Algorithms, Time-Travel, and GraphRAG in Pythonpy
- Graph Algorithms, Time-Travel, and GraphRAG in Rr
- See Your Graph: Visual Exploration with the Astraea UIboth
Walk
Embeddings, semantic search, and GraphRAG over a real corpus.
After this tier: You can turn a body of text into a knowledge graph and answer questions against it with an LLM grounded in what the graph actually says.
- What an Embedding Is, and How to Make Onepy
- Searching a Graph by Meaning, Not by Keywordpy
- Turning a Book Into a Knowledge Graphpy
- GraphRAG End to End: Retrieve, Extract, Linearize, Answerpy
- A Metadata Graph Over a Messy Data Lakepy
- Eunomia: A Semantic Cache in Front of Your Graphpy
- a-llama: A Local LLM Server That Rememberspy
Run
Investigative work on fraud and cyber graphs, and giving a coding assistant a memory.
After this tier: You can build an investigation graph from raw telemetry, hunt across it with embeddings and time travel, and produce a report where every claim traces back to a node.
Every lesson on this site was last run against AstraeaDB 9a9e919 on 2026-08-26. See the full verification status.