Technical Glossary

Isolated user-space process tree bounded by Linux namespaces and cgroups sharing the host kernel.

Immutable read-only tarball bundle containing root filesystem layers, binary dependencies, and OCI runtime metadata.

Linux kernel mechanism metering and limiting hardware resource allocations (CPU, memory, disk I/O) for process groups.

Kernel abstraction restricting what system resources (processes, network interfaces, mounts) a process tree can observe.

Declarative plain-text build manifest consumed by BuildKit to assemble container image layers sequentially.

Storage driver combining distinct read-only image layers and a read-write layer into a single unified directory tree.

Storage optimization strategy deferring file copies until modifications occur, sharing underlying layer blocks.

Concurrent, DAG-based image builder engine in Docker offering parallel step execution and secret mounting.

Dockerfile design pattern using multiple FROM instructions to separate build-time SDK dependencies from slim runtime artifacts.

Open Container Initiative specification defining container image tarball structures, manifests, and layer blobs.

CLI command delegating Dockerfile parsing and layer assembly to the BuildKit engine.

CLI command instantiating and launching a container with specified isolation boundaries, volumes, and ports.

CLI command spawning a new process inside the existing namespaces of a running container.

CLI command creating and starting multi-container services declared in a compose YAML manifest.

HTTP API service storing and serving versioned OCI image manifests and content-addressable layer blobs.

Smallest deployable unit in Kubernetes, grouping co-located containers sharing network namespaces and storage volumes.

Worker machine (physical or VM) running kubelet, container runtime, and kube-proxy in a Kubernetes cluster.

Declarative controller managing stateless Pod replicas, progressive rollouts, and automatic rollbacks.

Abstract network endpoint presenting a stable virtual IP and DNS name across a dynamic pool of Pods.

API object managing external HTTP/HTTPS routing rules to cluster-internal Services.

Continuous control loop pattern comparing actual cluster state against declared target state in etcd.

API object binding non-confidential key-value configuration data to Pod container environment variables or volume mounts.

API object storing base64-encoded confidential credentials, certificates, or keys separately from image code.

Workload controller managing stateful applications requiring unique network IDs and ordered volume provisioning.

CLI command declaratively updating cluster resources using 3-way strategic merge patches against etcd.

CLI command querying and displaying summary tables of specified Kubernetes API resources.

CLI command establishing interactive shell sessions inside a specific container in a cluster Pod.

Package manager for Kubernetes bundling YAML templates and default value overrides into versioned Charts.

Primary node agent ensuring containers described in PodSpecs are instantiated and healthy on worker hosts.

Neural network architecture relying entirely on self-attention mechanisms to compute contextual representations in parallel.

Mathematical operation computing context weights by comparing Query, Key, and Value vector dot-products.

Sub-word atomic chunk created by tokenizers (e.g. BPE) mapping text strings to numerical vocabulary IDs.

Dense numerical vector representing semantic meaning of text tokens in a continuous high-dimensional vector space.

Architecture pattern augmenting LLM prompts with relevant context retrieved dynamically from external vector databases.

Maximum token capacity an LLM can digest in a single forward pass, spanning system prompts, history, and responses.

Hyperparameter scaling logit probabilities before softmax to tune generation randomness and entropy.

Sampling strategy filtering candidate tokens to the smallest set whose cumulative probability reaches threshold P.

Process of updating LLM weights on specialized task datasets to adjust domain behavior or response formats.

Parameter-efficient fine-tuning technique freezing base weights and inserting trainable low-rank decomposition matrices.

Alignment technique tuning LLM outputs using reward models trained on human pairwise preference rankings.

Database engine specialized for indexing and performing fast nearest-neighbor similarity searches over vector embeddings.

Autonomous software loop leveraging an LLM as its core reasoning engine to observe environments, plan, and execute tool calls.

Mechanism enabling LLMs to emit structured JSON payload calls invoking external APIs, scripts, or database queries.

Agent prompting framework interleaving explicit step-by-step reasoning thoughts with action tool calls and observation inputs.

Control pattern decomposing high-level user tasks into DAG sub-goals executed sequentially or in parallel by subagents.

Agent self-correction loop evaluating tool outputs or generated code against criteria before returning final output.

Open client-server protocol standardizing how AI agents discover and invoke remote tools, resources, and prompts.

Architecture connecting multiple specialized subagents via a central router or swarm topology to solve complex tasks.

State persistence architecture dividing context into short-term working context, long-term vector memory, and episodic logs.