Architecture & AI Engineering Glossary.

A

API

Application Programming Interface. A set of rules and protocols that allows different software applications to communicate and exchange data. REST and GraphQL are common API styles used to expose functionality over HTTP.

API Gateway

A server that sits between clients and backend services, handling requests, authentication, rate limiting, and routing to the appropriate service. Simplifies client interactions with multiple microservices and centralizes cross-cutting concerns.

Architecture-first

A methodology that prioritizes system design and architecture decisions before writing code. Reduces costly refactoring, improves scalability from day one, and aligns engineering with business goals.

B

BaaS (Backend as a Service)

A cloud platform that provides server-side logic and database storage, eliminating the need to build and maintain your own backend infrastructure. Firebase and Supabase are popular examples.

C

CDN

Content Delivery Network. A distributed set of servers that caches and serves static content from locations near the user, reducing latency and improving page load times globally.

CI/CD

Continuous Integration and Continuous Deployment. Automated systems that test code on every commit and automatically deploy it to production when tests pass, enabling rapid and safe releases.

Code Review

A structured process where another engineer examines code before it merges, checking for bugs, architecture violations, and best practices. Critical quality gate and knowledge-sharing mechanism in professional teams.

Containerization

Packaging an application with all its dependencies into a self-contained unit called a container. Docker is the most common tool. Guarantees consistent behavior across development, staging, and production environments.

D

Domain-Driven Design

An architectural approach that models software around the business domain using shared language between engineers and domain experts. Aligns code structure with real-world business logic and improves long-term maintainability.

E

E-E-A-T

Experience, Expertise, Authoritativeness, Trustworthiness. Google's quality standard for content evaluation. Demonstrated through author credentials, bios, cited sources, and topical authority.

Event-Driven Architecture

A system design where components communicate by emitting and subscribing to events rather than direct calls. Decouples services, improves scalability, and enables real-time data propagation across systems.

F

FinOps

Financial Operations. Practices for optimizing cloud spending through monitoring, allocation, and governance. Prevents unexpected bills from unused resources and ensures cost efficiency at scale.

Function Calling

A capability of advanced LLMs like Claude and GPT-4 that allows the model to request execution of specific functions with parameters. Enables AI to interact with external tools and APIs rather than just generating text.

H

Headless CMS

A content management system that stores and manages content without a built-in presentation layer. Content is accessed via API, allowing any frontend to consume it.

HIPAA

Health Insurance Portability and Accountability Act. A US federal law regulating protected health information, in effect since 1996. Required compliance for any application storing or processing patient data.

I

Idempotency

A property where performing the same operation multiple times produces the same result as doing it once. Critical for payment processing, distributed systems, and API retries to prevent duplicate transactions.

K

KYC/AML

Know Your Customer and Anti-Money Laundering. Regulatory requirements in fintech requiring identity verification and ongoing monitoring for suspicious activity. Mandated by financial regulators globally.

L

LangChain

An open-source framework for building applications with LLMs. Provides abstractions for prompting, memory management, external tools, and chaining multiple AI calls together.

LangGraph

A state-machine framework for building complex AI agent workflows. Extends LangChain with graph-based control flow for multi-step agentic reasoning and decision-making.

LLM

Large Language Model. A neural network trained on vast text data to predict and generate human language. Claude, GPT-4, and Gemini are examples. Powers modern AI applications.

M

Microservice

A small, independently deployable service that handles one business capability. Multiple microservices combine via APIs to form a complete application, enabling independent scaling and deployment.

MVP

Minimum Viable Product. The smallest version of your product that solves the core user problem. Allows fast feedback and learning with minimal time and capital investment.

Multi-tenancy

An architecture where a single application instance serves multiple independent customers (tenants). Requires strict data isolation and careful permission modeling to prevent cross-tenant data exposure.

O

Observability

The ability to understand system behavior through logs, metrics, and traces. Enables rapid debugging and performance optimization in production without guessing what went wrong.

OpenAPI

A standard format for describing REST APIs using JSON or YAML. Enables automatic API documentation, client code generation, and testing tools.

P

PCI-DSS

Payment Card Industry Data Security Standard. A compliance framework for systems handling credit card data, requiring encryption, access controls, and regular audits.

R

RAG (Retrieval-Augmented Generation)

An AI technique that retrieves relevant documents before the model generates a response, reducing hallucinations and providing current, sourced information instead of relying solely on training data.

Rate Limiting

A technique that restricts the number of requests a user or service can make within a time window. Protects against abuse, prevents cascading failures, and ensures fair resource allocation.

RDBMS

Relational Database Management System. A database that organizes data into tables with relationships between them. PostgreSQL and MySQL are standard production choices.

RLS (Row-Level Security)

A database feature that automatically filters which rows a user can see based on their identity and permissions. Essential for multi-tenant systems and enforcing data isolation at the database layer.

S

Schema

The structure and organization of data in a database or API. Defines what fields exist, their types, and relationships. Acts as a contract between data producer and consumer.

SDLC

Software Development Life Cycle. The complete process from planning through design, development, testing, deployment, and maintenance.

Serverless

A cloud model where you write functions that execute on-demand without managing servers. AWS Lambda and Google Cloud Functions are examples. You pay only for actual execution time.

SLA

Service Level Agreement. A contract specifying availability, performance, and support guarantees. Example: 99.9 percent uptime with 1-hour response time for critical issues.

SLO

Service Level Objective. Internal targets for system performance and reliability. Example: p99 latency under 100ms. Used to guide engineering investment decisions.

SSE (Server-Sent Events)

A protocol allowing servers to push data to clients over HTTP. Used for real-time updates like live notifications or streaming AI responses without websockets.

SSR (Server-Side Rendering)

Rendering a web page on the server before sending HTML to the browser. Improves SEO and initial load time compared to client-side rendering.

Stripe Connect

Stripe's platform for marketplaces and platforms to accept payments on behalf of users. Handles payouts, identity verification, and regulatory compliance automatically.

Strangler Fig Pattern

A safe refactoring technique where new code gradually replaces legacy code. Parallel systems coexist until migration is complete, minimizing risk and allowing rollback.

T

Tool Use

The ability of an AI model to call external functions and APIs. Extends the model beyond text generation to interact with real systems, databases, and tools.

V

Vector Database

A database optimized for storing and searching high-dimensional vectors. Essential for semantic search and RAG systems. Pinecone and Weaviate are examples.

W

Webhook

An HTTP callback that a service triggers when an event occurs, notifying another system immediately. Enables real-time event-driven integration without polling.

Z

Zero-Downtime Deployment

Releasing new code without taking the service offline. Uses blue-green deployments, load balancers, and database migrations that maintain availability during transitions.