AI & Data Glossary
Clear, concise definitions of key AI, machine learning, and data analytics terms.
RAG (Retrieval-Augmented Generation)
RAG is a technique that combines large language models with external knowledge retrieval. Instead of relying solely on the model's training data, RAG retrieves relevant documents from a knowledge base and passes them as context to the LLM, enabling accurate, up-to-date, and source-cited answers.
LLM (Large Language Model)
An LLM is an AI model trained on vast amounts of text data to understand and generate human-like text. Examples include GPT-4, Claude, and Llama. LLMs power chatbots, content generation, document processing, and code assistance.
MLOps (Machine Learning Operations)
MLOps is the practice of deploying, monitoring, and maintaining machine learning models in production. It encompasses CI/CD pipelines for models, performance monitoring, drift detection, automated retraining, and version control for reproducibility.
Fine-Tuning
Fine-tuning is the process of taking a pre-trained model and further training it on domain-specific data to adjust its weights. This is ideal for achieving consistent style, tone, or specialized behavior, but requires labeled data and compute resources.
Prompt Engineering
Prompt engineering is the practice of designing input prompts to elicit desired outputs from LLMs. Techniques include few-shot examples, chain-of-thought reasoning, and structured output formatting to improve accuracy, consistency, and control.
Zero-Shot Prompting
Zero-shot prompting is asking an LLM to perform a task without providing any examples in the prompt. The model relies entirely on its pre-trained knowledge to understand the instruction and generate a response.
Few-Shot Prompting
Few-shot prompting is providing an LLM with 1-5 examples of the desired input-output pattern before asking it to perform the task. This improves accuracy and consistency by demonstrating the expected format and reasoning.
Chain-of-Thought
Chain-of-thought is a prompting technique that asks an LLM to show its reasoning step-by-step before giving a final answer. This improves accuracy on complex reasoning tasks like math, logic, and multi-step analysis.
Embeddings
Embeddings are numerical vector representations of text that capture semantic meaning. They enable similarity search, clustering, and retrieval by converting words or documents into arrays of numbers that machine learning models can compare mathematically.
Vector Database
A vector database is a specialized database designed to store and query embedding vectors. It enables fast similarity search, making it the backbone of RAG systems. Popular options include Pinecone, Weaviate, and Chroma.
Hallucination
Hallucination occurs when an LLM generates confident but factually incorrect information. It happens when the model lacks specific knowledge but generates plausible-sounding content instead of acknowledging uncertainty. RAG and guardrails reduce hallucination.
Reranking
Reranking is a two-stage retrieval technique in RAG systems. First, a fast retriever fetches candidate documents via vector similarity, then a slower but more accurate cross-encoder model re-scores them, improving precision.
Guardrails
Guardrails are safety mechanisms in AI systems that validate inputs, filter outputs, and prevent failures. They include query validation, prompt injection detection, relevance scoring, fact verification, and hallucination detection.
CRAG (Corrective RAG)
CRAG is an advanced RAG technique that adds a self-correction mechanism. Instead of blindly using retrieved documents, CRAG evaluates their relevance and quality before passing them to the LLM, filtering irrelevant content to reduce hallucinations.
Transfer Learning
Transfer learning is reusing a model trained on one task as the starting point for a different but related task. This dramatically reduces the data and compute needed, as the model retains general knowledge from its original training.
Data Warehouse
A data warehouse is a centralized repository that consolidates data from multiple sources for analytics and reporting. Unlike operational databases, it's optimized for read-heavy analytical queries across large historical datasets.
ETL/ELT
ETL (Extract, Transform, Load) and ELT (Extract, Load, Transform) are data pipeline patterns. ETL transforms data before loading; ELT loads raw data first and transforms it in the warehouse using its compute power.
Business Intelligence (BI)
BI is the practice of collecting, analyzing, and visualizing business data to inform decision-making. It includes dashboards, reports, and analytics platforms that transform raw data into insights accessible to non-technical stakeholders.
Related Articles
Have Questions About These Terms?
Our team can help you understand how these concepts apply to your business.
Ask Our Experts