Vector Database

In AI, hallucination refers to when a language model generates confident-sounding but factually incorrect or fabricated information. This occurs because LLMs predict statistically likely text rather than retrieving verified facts. Mitigation strategies include RAG, grounding responses in source documents, structured output validation, and using temperature settings to reduce creative deviation.

A context window is the maximum amount of text (measured in tokens) that an LLM can process in a single interaction, encompassing both the input prompt and the generated output. Larger context windows allow models to handle longer documents, maintain extended conversations, and reason over more information at once. Context window sizes have grown rapidly — from 4K tokens in early GPT models to over 1M tokens in current models like Claude.

Chain of Thought (CoT) is a prompting technique that encourages an LLM to break down complex reasoning into intermediate steps before arriving at a final answer. By explicitly reasoning through each step, models achieve significantly better accuracy on math, logic, and multi-step problems. Extended thinking and "thinking" tokens in models like Claude represent a built-in form of chain-of-thought reasoning.

ETL (Extract, Transform, Load) is an automated data processing pattern where data is extracted from source systems, transformed into a desired format or structure, and loaded into a target system like a data warehouse. Modern variations include ELT, where raw data is loaded first and transformed in place. ETL pipelines are essential for automating data integration, reporting, and feeding clean data into ML training workflows.

An embedding is a dense numerical vector representation of data — such as text, images, or code — in a high-dimensional space where semantically similar items are positioned closer together. Embeddings are fundamental to semantic search, recommendation systems, and RAG pipelines. They are generated by specialized models and typically stored in vector databases for efficient similarity lookups.

Retrieval-Augmented Generation (RAG) is a technique that enhances LLM responses by retrieving relevant documents from an external knowledge base before generating an answer. This allows the model to ground its output in up-to-date, domain-specific information rather than relying solely on its training data. RAG is widely used in enterprise chatbots, documentation assistants, and search-powered AI applications.