AI Engineer Roadmap
From LLM fundamentals and prompt engineering through embeddings, retrieval-augmented generation, tool use, and evaluation, the core path for building real LLM-powered applications, linked into Cloud Tech by Victor topic references.
Drag to pan · Tab through steps · arrow keys jump between connected steps
Interactive diagram. Press Tab to move through the 6 steps in order; each step links to its full reference page.
Tokens, context windows, and sampling parameters are the foundation everything else in LLM application development builds on.
System prompts and well-chosen examples are the most reliable levers for controlling an LLM's behavior and output format.
Embeddings and cosine similarity are what make semantic search, and everything built on top of retrieval, possible.
RAG retrieves only what's relevant at request time, and how chunks are prepared before embedding decides whether retrieval actually works.
Tool use turns a text generator into an agent that can call real functions and act on their results.
An automated eval suite and concrete hallucination-reduction techniques are what turn 'it feels better' into a shippable decision.