Clear, no-jargon definitions of the concepts behind our courses — from containers and Kubernetes to RAG, agents and MCP. Each term links to a full explainer and the course that teaches it.
The ability of an AI agent to persist information across separate runs — recalling past context instead of starting blank each time — which is what makes long-running, resumable workflows possible.
An AI system that does more than answer — it decides on and takes actions using tools (reading systems, searching docs, running commands) to accomplish a goal, usually in a loop.
An Application Programming Interface: a defined contract that lets one piece of software talk to another, exchanging structured requests and responses rather than screens meant for humans.
Continuous Integration and Continuous Delivery/Deployment: the practice of automatically building, testing, and shipping every code change through a pipeline, instead of releasing by hand.
The text interface (shell) where you type commands to control a computer directly — the foundational tool for Linux, servers, and every deployment workflow.
Kubernetes objects for supplying configuration to containers — ConfigMaps for non-sensitive settings, Secrets for sensitive values like passwords and tokens — kept separate from the image.
A lightweight, isolated package of an application and everything it needs to run, sharing the host’s kernel — not a full virtual machine, which makes it fast to start and cheap to run.
The Domain Name System: the internet’s directory that translates a human-friendly domain name into the IP address a browser actually connects to.
Named values set outside your code and read at runtime — the standard way to pass configuration and secrets to a program without hard-coding them.
The number a program returns when it finishes: 0 means success, anything else signals a specific failure — the primary way scripts and orchestrators know whether a step worked.
The Linux rules that control who can read, write, or execute a file or directory — the model behind most "permission denied" errors and a core part of server security.
A commit is a saved snapshot of your project; a branch is an independent line of commits. Together they let you version history and work on changes in isolation before merging.
HTTPS is HTTP encrypted with TLS, so traffic between browser and server is private and tamper-proof and the server’s identity is verified by a certificate.
Defining servers, networks, and services in version-controlled files instead of clicking through consoles — so environments are repeatable, reviewable, and rebuildable.
An open-source system for running containers across many machines — scheduling them, restarting failures, scaling, and networking them together so applications self-heal and stay available.
A stable network endpoint that routes traffic to a changing set of pods. ClusterIP exposes it inside the cluster, NodePort and LoadBalancer expose it outside.
An open protocol that lets AI applications connect to external tools and data through a standard interface, so one server of tools can be used by any compatible client.
Several specialised AI agents coordinating on a task — each with a focused role and limited tools — instead of one agent trying to do everything, enabling separation of duties and safer autonomy.
A way of structuring code around objects — bundles of data and the methods that act on it — using classes and inheritance to build software that scales beyond a single script.
A Pod is the smallest deployable unit in Kubernetes (one or more containers); a ReplicaSet keeps a set number of identical pods running; a Deployment manages ReplicaSets to roll out updates safely.
What actually happens when you run a program — source is interpreted or compiled, loaded into memory as a process, and executed by the CPU — the mental model behind most runtime errors.
An attack where malicious text hidden in an AI system’s input hijacks its instructions — the core security risk to understand before letting an AI agent read untrusted data or take actions.
The Prometheus Query Language, used to turn raw time-series metrics into rates, aggregations, and alerts — the language behind every real monitoring dashboard.
A technique that retrieves relevant documents and feeds them to a language model so its answers are grounded in your data and can be cited — rather than made up from training alone.
A server that sits in front of your applications, receiving client requests and forwarding them on — handling TLS, routing, and load balancing in one place. Nginx is the classic example.
A cryptographic key pair — a public key you share and a private key you keep — used to log in to servers and services securely without passwords.
The mechanism (also called function calling) that lets a language model invoke real functions — checking a system, querying a database, sending a request — turning a chatbot into an agent that can act.
Numeric representations of text (or other data) that place similar meanings close together in space, enabling semantic search — the retrieval half of a RAG system.
Isolated per-project Python installations that keep each project’s dependencies separate, so packages for one project never conflict with another.
Every term here is something you build with, hands-on, in a BytExplorer course.