Engineering & AI glossary
Understand the metrics, models, and methods behind modern engineering. Clear definitions, practical examples, and a closer look at what the numbers actually mean.
All terms
2,008 termsTool capability routing
Tool capability routing is a model-routing or gateway concept used to manage operational visibility for AI requests. It describes a distinct decision, control, interface, or observation point between an application and one or more model providers.
Model routing and gatewaysTool latency
Tool latency is the time an application or agent waits for a tool call to complete. It includes work such as a database query, API request, file operation, or test run between model responses.
Inference performanceTool-use evaluation
Tool-use evaluation is a defined lens for examining AI system behavior with a stated task, evidence, and interpretation rule.
Evaluations and benchmarksTop-k sampling
Top-k sampling is a language-model concept about representation and similarity. It names a mechanism, representation, prompting pattern, decoding control, or context behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTop-p sampling
Top-p sampling, also called nucleus sampling, selects the next token from the smallest group whose cumulative probability reaches a chosen probability p. Lower p values restrict the candidate set, while higher values allow more candidates.
LLM fundamentalsToxicity evaluation
Toxicity evaluation is a defined lens for examining AI system behavior with a stated task, evidence, and interpretation rule.
Evaluations and benchmarksTrace context
Trace context is the small set of identifiers and flags passed between components so telemetry can associate an operation with its trace and parent span. It allows distributed work to be represented as one connected execution.
Reliability and observabilityTrace ID
Trace ID is the identifier shared by spans belonging to one distributed operation.
Reliability and observabilityTrace sampling
Trace sampling is selection of traces or spans to retain from a larger set of observed operations.
Reliability and observabilityTrace signal
Trace signal is evidence that records the path and timing of one operation across services, queues, and databases.
Reliability and observabilityTrace waterfall
Trace waterfall is a time-aligned view showing span nesting, overlap, waiting, and duration.
Reliability and observabilityTraffic mirroring
Traffic mirroring is a model-routing or gateway concept used to manage traffic selection for AI requests. It describes a distinct decision, control, interface, or observation point between an application and one or more model providers.
Model routing and gatewaysTraffic shadowing
Traffic shadowing is a release engineering and DevOps concept for controlling how software changes are prepared, introduced, or understood.
DORA and DevOpsTrain-test contamination
Train-test contamination is a defined lens for examining AI system behavior with a stated task, evidence, and interpretation rule.
Evaluations and benchmarksTraining batch size
Batch size is a language-model concept about serving behavior and operational tradeoffs. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTraining data contamination
Training data contamination is a language-model concept about training behavior and measurement. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTraining loss
Training loss is a language-model concept about training behavior and measurement. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTraining step
Training step is a language-model concept about training behavior and measurement. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTransformer block
Transformer block is a language-model concept about mechanism and information flow. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTransitive dependency
Transitive dependency is a software maintenance concern describing a condition that can make future changes, verification, operation, or ownership harder. Its practical importance depends on supported behavior, rate of change, and the consequences of delay.
Code quality and technical debtTreatment effect
Treatment effect is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationTreatment group
Treatment group is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationTreatment spillover
Treatment spillover is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationTree of thoughts
Tree of thoughts is a language-model concept about instruction design and control. It names a mechanism, representation, prompting pattern, decoding control, or context behavior that can change how an AI system processes input and produces output.
LLM fundamentalsTriage queue
A triage queue contains requests that have arrived but have not yet been classified or routed into a delivery, incident, support, or rejection path. Its age measures intake delay rather than implementation delay.
Flow and capacity planningTrunk stability
Trunk stability is a software delivery concept used to describe a specific event, interval, control, or operating condition in the path from source change to production behavior. A useful definition names the boundary, unit, and decision the measure supports.
DORA and DevOpsTrunk-based development
Trunk-based development is a version control practice where developers integrate small changes into a shared main branch frequently. Short-lived branches and feature flags can keep incomplete work safe while reducing long-lived merge divergence.
DORA and DevOpsTrunk-based release
Trunk-based release is a release engineering and DevOps concept for controlling how software changes are prepared, introduced, or understood.
DORA and DevOpsType i error
Type i error is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationType ii error
Type ii error is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationTypical sampling
Typical sampling is a language-model concept about context selection and limits. It names a mechanism, representation, prompting pattern, decoding control, or context behavior that can change how an AI system processes input and produces output.
LLM fundamentalsUncached token
Uncached token is an input token processed without a reusable cache hit. It gives teams a way to name, measure, or reason about an economic property of an AI workload without treating raw usage as proof of value.
Token costs and AI ROIUnderfitting
Underfitting is a language-model concept about mechanism and information flow. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsUnified model API
Unified model API is a model-routing or gateway concept used to manage interface consistency for AI requests. It describes a distinct decision, control, interface, or observation point between an application and one or more model providers.
Model routing and gatewaysUnique identifier
Unique identifier distinguishes an event, entity, or work item in scope.
Engineering analyticsUnit of analysis
Unit of analysis is a statistical or measurement concept used to describe, compare, or interpret engineering data. Its meaning depends on the unit of analysis, data-generating process, and question being asked.
Measurement and experimentationUnit of measure
Unit of measure states what a metric value represents.
Engineering analyticsUnit testing
Unit testing is the practice of checking a small, replaceable part of a program in isolation from its collaborators. A unit test usually supplies controlled inputs, observes behavior, and compares the result with an expected outcome.
Code quality and technical debtUnplanned work
Unplanned work is a developer productivity concept that helps teams understand unplanned work in the context of software delivery.
Developer productivityUnplanned work rate
Unplanned work rate is a developer productivity concept that helps teams understand unplanned work rate in the context of software delivery.
Developer productivityUpstream dependency
Upstream dependency is a concept used in software delivery planning to describe a condition, relationship, estimate, or decision about engineering work.
Flow and capacity planningUrgency
Urgency is a concept used in software delivery planning to describe a condition, relationship, estimate, or decision about engineering work.
Flow and capacity planningUsability debt
Usability debt is a software maintenance concern describing a condition that can make future changes, verification, operation, or ownership harder. Its practical importance depends on supported behavior, rate of change, and the consequences of delay.
Code quality and technical debtUsability testing
Usability testing is a software testing or test-design practice used to gather evidence about a defined risk, behavior, boundary, or operating condition. It makes the question under test explicit, identifies the inputs and observations that matter, and gives a team a repeatable basis for deciding whether the result is acceptable.
Code quality and technical debtUsage normalization
Usage normalization is a model-routing or gateway concept used to manage interface consistency for AI requests. It describes a distinct decision, control, interface, or observation point between an application and one or more model providers.
Model routing and gatewaysUse case testing
Use case testing is a software testing or test-design practice used to gather evidence about a defined risk, behavior, boundary, or operating condition. It makes the question under test explicit, identifies the inputs and observations that matter, and gives a team a repeatable basis for deciding whether the result is acceptable.
Code quality and technical debtUser research
User research is a way to organize, support, or evaluate software work so that teams can make useful progress with less avoidable friction. It is most valuable when connected to a concrete outcome and the local conditions of the team using it.
Developer productivityUser-visible error
User-visible error is a reliability concept used to describe a specific condition, control, or decision in the operation of software services.
Reliability and observabilityValidation loss
Validation loss is a language-model concept about serving behavior and operational tradeoffs. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsValidation set
Validation set is a defined lens for examining AI system behavior with a stated task, evidence, and interpretation rule.
Evaluations and benchmarks