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 termsSoftware factory baseline
A software factory baseline is a documented snapshot of a software delivery system before an intervention or comparison. It records the work population, event definitions, time window, measures, and relevant operating conditions needed to interpret later change.
Engineering analyticsSoftware factory bottleneck
A software factory bottleneck is a person, team, policy, tool, or workflow stage whose effective capacity limits the rate at which the delivery system can complete work. It is identified through sustained evidence of constrained flow and queues.
Flow and capacity planningSoftware factory cadence
Software factory cadence is the recurring rhythm by which a software delivery system receives work, creates changes, gathers feedback, releases software, and reviews outcomes. It is a property of the system's flow, not a requirement that every team work to the same schedule.
Engineering analyticsSoftware factory capacity
Software factory capacity is the amount of work a software delivery system can complete during a defined period under stated conditions. It depends on people, tools, queues, policies, work mix, dependencies, and quality requirements, so it is not a fixed count of engineers multiplied by hours.
Flow and capacity planningSoftware factory change size
Software factory change size is the amount of code or delivery scope included in a software change as defined by a chosen unit. It is a context signal that can affect review effort, feedback speed, failure isolation, and queue behavior.
Developer productivitySoftware factory feedback loop
A software factory feedback loop is a recurring path in which an observation about a software change or outcome informs a decision, the decision changes the system, and a later observation tests the result. Feedback may come from code review, tests, delivery, production, customers, or developers.
Engineering analyticsSoftware factory governance
Software factory governance is the set of policies, controls, ownership rules, and review practices used to guide software delivery toward security, reliability, compliance, and product goals. Effective governance makes expectations visible and provides a workable path for exceptions.
DORA and DevOpsSoftware factory improvement experiment
A software factory improvement experiment is a time-bounded change to a software delivery system that tests whether a specific intervention improves a defined outcome. It uses a baseline, a stated hypothesis, comparable measures, and a review of the evidence before the change is adopted more broadly.
Measurement and experimentationSoftware factory metrics
Software factory metrics are a defined set of measurements used to understand how an organization's software delivery system performs. They cover the movement, quality, stability, cost, and experience of work rather than reducing the factory to one activity count.
Measurement and experimentationSoftware factory observability
Software factory observability is the ability to understand what is happening inside a software delivery system by using connected signals about work, workflow state, timing, failures, ownership, and outcomes. It supports investigation by preserving enough context to explain why a result occurred.
Engineering analyticsSoftware factory queue time
Software factory queue time is the elapsed time a software work item spends waiting for a person, decision, resource, check, environment, or next workflow stage. It is a part of total delivery time and should be defined by the queue boundary being measured.
Engineering analyticsSoftware factory rework
Software factory rework is software work performed again because an earlier change was defective, incomplete, misunderstood, rejected, or made obsolete. It includes corrective changes and repeated effort that consumes delivery capacity without representing a new independent outcome.
Engineering analyticsSoftware factory stability
Software factory stability is the ability of a software delivery system to release changes predictably, limit the harm from failures, and recover when intervention is required. It describes the behavior of the system around delivery, rather than the absence of all change or risk.
DORA and DevOpsSoftware factory throughput
Software factory throughput is the amount of software work that reaches an agreed completion point during a defined period. The completion point may be a merged change, a production deployment, or a customer outcome, and the chosen boundary must remain consistent.
Engineering analyticsSoftware factory work in progress
Software factory work in progress is the unfinished software work currently carried by a delivery system. It includes changes being implemented as well as work waiting for review, testing, approval, deployment, clarification, or another dependency.
Flow and capacity planningSoftware supply chain
Software supply chain is a release engineering and DevOps concept for controlling how software changes are prepared, introduced, or understood.
DORA and DevOpsSoftware value stream
A software value stream is the sequence of activities and handoffs required to turn a software request into a usable outcome. It includes the work that creates new value and the recovery work required to restore or protect an existing service.
Engineering analyticsSource system
Source system identifies where measurement records originate.
Engineering analyticsSPACE framework
The SPACE framework is a multidimensional approach to understanding developer productivity across Satisfaction and well-being, Performance, Activity, Communication and collaboration, and Efficiency and flow.
Developer productivitySpan
Span is a timed unit of work within a trace with operation, timing, status, attributes, and events.
Reliability and observabilitySpan attribute
Span attribute is a key-value property attached to a span to provide searchable operation context.
Reliability and observabilitySpan event
Span event is a timestamped annotation attached to a span to record something during its lifetime.
Reliability and observabilitySpan ID
Span ID is the identifier for one span within a trace that distinguishes it from its parent and siblings.
Reliability and observabilitySparse autoencoder
Sparse autoencoder is a language-model concept about evaluation design and failure analysis. It names a mechanism, representation, training practice, or operational behavior that can change how an AI system processes input and produces output.
LLM fundamentalsSpecial cause variation
Special cause variation 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 experimentationSpecial token
Special token 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 fundamentalsSpecialist queue
A specialist queue is a set of items that require capability available from only a small number of people or systems. It is a queue caused by constrained expertise rather than by total team capacity alone.
Flow and capacity planningSpecification by example
Specification by example 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 debtSpeculative decoding
Speculative decoding is the serving concept concerned with speculative decoding during AI inference.
Inference performanceSpeculative generality
A code smell in which abstractions or extension points exist for hypothetical future needs.
Code quality and technical debtSpend attribution
Spend attribution is the mapping of AI spend to a product, team, workflow, user, or outcome. 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 ROISpend limit
Spend limit is a model-routing or gateway concept used to manage policy enforcement 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 gatewaysSpike testing
Spike 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 debtSplit brain
Split brain is a reliability concept used to describe a specific condition, control, or decision in the operation of software services.
Reliability and observabilitySprint planning
Sprint planning is a concept used in software delivery planning to describe a condition, relationship, estimate, or decision about engineering work.
Flow and capacity planningSquash merge
Squash merge is a practical concept in a pull-request workflow that shapes how people examine, discuss, own, or integrate a proposed change.
Code reviewStacked pull requests
Stacked pull requests divide dependent work into a sequence of smaller changes that can be reviewed in order.
Code reviewStaged rollout
Staged rollout is a release engineering and DevOps concept for controlling how software changes are prepared, introduced, or understood.
DORA and DevOpsStakeholder alignment
Stakeholder alignment 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 productivityStale review
Stale review is a practical concept in a pull-request workflow that shapes how people examine, discuss, own, or integrate a proposed change.
Code reviewStandard change
Standard change is a release engineering and DevOps concept for controlling how software changes are prepared, introduced, or understood.
DORA and DevOpsStandard error
Standard 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 experimentationStandard item
A standard item is work that follows the ordinary pull order, WIP policy, and service expectation of a workflow. It is distinguished from urgent, fixed-date, or other special classes of service.
Flow and capacity planningStartup probe
Startup probe is a reliability concept used to describe a specific condition, control, or decision in the operation of software services.
Reliability and observabilityState transition testing
State transition 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 debtStatic batching
Static batching is the serving concept concerned with static batching during AI inference.
Inference performanceStatic routing
Static routing 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 gatewaysStatic-analysis false-positive rate
The share of reported static-analysis findings that reviewers determine do not represent actionable issues.
Code quality and technical debtStationarity
Stationarity 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 experimentationStatistical power
Statistical power is a defined lens for examining AI system behavior with a stated task, evidence, and interpretation rule.
Evaluations and benchmarks