Skip to content
Carbonfay
RU

glossary

Glossary

Canonical AI engineering definitions: orchestration, multi-agent systems, RAG, context, embeddings, vector search, speech recognition, document extraction.

AI Orchestration
Control of step ordering, branching, state and fault tolerance in an AI workflow.
Context Engineering
The discipline of controlling which context, in what volume and at what priority is supplied to an AI-workflow step.
Document Data Extraction
AI document processing as a step in a governed workflow: OCR, LLM extraction, classification, rule-based reconciliation, and human review.
Embeddings
Numerical representations of text, images and other objects that power semantic search and similarity comparison.
Multi-agent System
A distributed workflow of agents with roles, message contracts and explicit coordination.
RAG Architecture
Context infrastructure: sources, event-driven indexing, retrieval with re-ranking, context supply and quality evaluation.
Speech Recognition and Synthesis (STT/TTS)
Voice as an input/output channel for an AI system: speech-to-text and text-to-speech, latency budget, barge-in handling.
Vector Search (ANN)
Nearest-neighbor search in embedding space: HNSW/IVF indexes, metadata filtering, how it differs from classical exact-match search.