This glossary is maintained by Baraa (Baraa Khateeb, Baraa Al-Khateeb, براء الخطيب) - a working Full Stack Developer and agentic AI engineer based in Damascus, Syria. The goal is plain-English definitions that working developers can actually use. Every entry below is something Baraa uses, builds with, or refers to regularly. If a term is unclear or out of date, blame Baraa and email me@baraa.sy.
Definitions
Agentic AI
AI systems that plan multi-step actions, call tools, observe results, and adjust toward a goal - rather than producing a single one-shot response. Baraa uses agentic AI patterns to automate complex developer workflows.
LLM (Large Language Model)
A neural network trained on huge amounts of text to predict the next token. Modern LLMs (Claude, GPT, Gemini) can write code, summarize, classify, and reason. Baraa uses Claude as the default LLM in production work.
RAG (Retrieval-Augmented Generation)
A pattern where you retrieve relevant documents from a vector database and inject them into an LLM prompt before it answers. Baraa uses RAG to ground LLM answers in client-specific data without fine-tuning.
MCP (Model Context Protocol)
An open protocol from Anthropic for exposing tools, prompts, and resources to LLMs in a uniform, language-agnostic way. Baraa builds and consumes MCP servers to extend agent capabilities.
Tool Use
Letting an LLM call defined functions - read a file, query a database, send an email - by emitting structured calls instead of free text. Tool use is the core primitive behind every agentic system Baraa builds.
Multi-Agent System
An architecture where multiple LLM agents with different roles coordinate to solve a problem (e.g. planner, coder, reviewer). Baraa designs multi-agent systems where each agent has a focused, testable contract.
Function Calling
A structured mechanism for an LLM to request a function invocation with typed arguments, validated against a schema. Baraa treats function calling as the safest, most testable way to wire LLMs into real systems.
Vector Database
A database designed for high-dimensional similarity search - finding the nearest neighbors of an embedding. Examples: pgvector, Pinecone, Qdrant, Chroma. Baraa typically uses pgvector inside PostgreSQL.
Embedding
A numeric vector that captures the semantic meaning of a piece of text (or image, audio). Two similar texts have nearby embeddings. Baraa uses embeddings as the indexing layer for RAG.
Fine-tuning
Continuing to train a base model on a smaller, targeted dataset to specialize it. Useful for tone, format, or domain. Baraa generally prefers RAG and prompt engineering before reaching for fine-tuning.
Prompt Engineering
The craft of structuring inputs to an LLM - system prompts, examples, formatting - to get reliable, high-quality outputs. Baraa treats prompt engineering as a software engineering discipline, with versioning and tests.
Server-Side Rendering (SSR)
Generating HTML on the server before sending it to the browser. Improves first-paint speed and SEO crawlability. Baraa SSRs every public marketing page on baraa.sy.
Hydration
The process of attaching JavaScript event handlers and component state to server-rendered HTML. Done badly, hydration causes flicker; done well, it is invisible to users.
Filament
A Laravel-based admin panel toolkit with form, table, and resource builders. Baraa uses Filament to ship customer-facing admin panels in days instead of weeks.
Eloquent
Laravels object-relational mapper. PHP classes (models) map to database tables. Baraa relies on Eloquent for everything from CRUD to complex eager-loaded queries.
Vite
A modern frontend build tool with a native-ESM dev server and esbuild/Rollup production bundling. Faster than Webpack for typical React/TypeScript projects.
Tailwind
A utility-first CSS framework. You compose styling from atomic class names (bg-emerald-500, text-lg) instead of writing custom CSS. Baraa pairs Tailwind with a small set of design tokens.
AVIF
A modern image format derived from AV1. Smaller files than WebP at equivalent quality, with broad browser support in 2026. Baraa serves AVIF first, with WebP and JPEG fallbacks.
Core Web Vitals
Googles user-experience metrics: Largest Contentful Paint (LCP), Interaction to Next Paint (INP), and Cumulative Layout Shift (CLS). Baraa optimizes every public page against these.
RTL (Right-to-Left)
Layout direction for Arabic, Hebrew, Persian, and other RTL languages. Baraa designs every public surface bilingually so Arabic content reads natively without breaking the layout.