Explore the interactive map of the field, then pick a guided path. New to AI infrastructure? Start at the beginning. Already shipping? Jump into a deep dive. Everything here is written for builders who want to actually use AI — not just read about it.
Drag nodes, filter by domain, and click any concept to see what it means and how it connects.
Tip: drag nodes to rearrange the graph, hover to highlight neighbours, and use the search box to jump to any concept.
The same structure that makes this map readable is what gives production AI agents durable, explainable memory.
Vector search finds similar text; a knowledge graph stores facts and relationships an agent can traverse across sessions — the foundation of long-term agent memory.
GraphRAG lets a model follow chains of relationships to answer questions flat retrieval can't — "who depends on the service that owns this dataset?"
Answers can cite the exact entities and edges they came from, cutting hallucination and giving you an audit trail regulators actually accept.
Pick where you are. Each path is an ordered sequence of guides, concepts, and hands-on references.
Zero jargon. Understand what it actually takes to run an AI model in the real world.
Go from a single prompt to an agent that plans, uses tools, and remembers — reliably.
The practitioner's toolbox for making LLM serving dramatically cheaper and faster.
The honest map of everything between a model that works on your laptop and one that serves real users reliably.
KV-cache reuse, speculative decoding, prompt compression, and continuous batching — with an open-source stack.
Train → evaluate → gate → canary → roll out, with zero downtime — using DVC, MLflow, ArgoCD, and friends.
Why structured memory beats raw vector search for reasoning agents — and how GraphRAG actually works.
A curated digest of what's moving in production AI. Filter by topic.
The fastest way to learn this stack is to ship it alongside an expert. Book a free session and bring your hardest problem.