generated: '2026-09-06'
method: derived
source: https://studio.edgeimpulse.com/openapi.yml (481 paths, 670 component schemas) + openapi/edge-impulse-*-openapi.yml
provider: Edge Impulse
providerId: edge-impulse
description: 'Entity-relationship graph derived from the Edge Impulse Studio OpenAPI — path templates
  and id-reference fields across the 670 component schemas. The model is a two-root hierarchy: an Organization
  (Enterprise) owns Projects, and a Project owns everything else. 275 of 481 paths are scoped by {projectId}
  and 148 by {organizationId}; almost nothing exists outside one of those two roots, which is why a project-scoped
  API key is a genuine blast-radius boundary and a user JWT is not.'
id_conventions:
  style: integer surrogate keys
  note: Ids are plain integers, not prefixed strings — projectId 12345, not proj_12345. There is no type
    information in an id, so an agent cannot tell a projectId from an organizationId by looking at it.
    The one exception is the job id, a string of the form job-1569583053767.
entities:
- name: Organization
  key: organizationId
  root: true
  scope: Enterprise only
  description: 'Enterprise tenancy root: members, buckets, datasets, pipelines, portals, custom blocks,
    white labels.'
  paths: 148
- name: Project
  key: projectId
  root: true
  description: The primary workspace. Owns data, impulses, models, deployments, devices, jobs and keys.
  paths: 275
- name: User
  key: userId
  description: An account. Belongs to organizations and collaborates on projects.
- name: Sample
  key: sampleId
  description: One item of raw training, validation, testing or post-processing data, with labels and
    metadata.
  paths: 38
- name: Impulse
  key: impulseId
  description: 'The ML pipeline: input block plus processing (DSP) blocks plus learning blocks.'
- name: DSPBlock
  key: dspId
  description: A processing/feature-extraction block inside an impulse.
  paths: 20
- name: LearnBlock
  key: learnId
  description: A learning block (classification, regression, anomaly, object detection, transfer learning).
  paths: 19
- name: Job
  key: jobId
  id_format: job-<epoch-millis>
  description: An asynchronous unit of work — feature generation, training, evaluation, deployment build,
    export.
  paths: 16
- name: Device
  key: deviceId
  description: A connected edge device registered to a project.
  paths: 9
- name: Deployment
  description: A built on-device artifact for a specific deployment target, now with a version history.
- name: ProjectVersion
  key: versionId
  description: A snapshot of a project. The only restore point the platform offers.
- name: ApiKey
  key: apiKeyId
  description: A project-scoped credential (ei_ prefix). Ingestion-only variants exist.
- name: HmacKey
  key: hmacId
  description: A project-scoped signing key for ingestion payloads.
- name: Bucket
  key: bucketId
  description: An organization storage bucket (S3/GCS) backing organization datasets.
- name: OrganizationDataset
  key: dataset
  description: An organization-level dataset, distinct from project data.
  paths: 11
- name: DataItem
  key: dataId
  description: One item inside an organization dataset.
  paths: 6
- name: Pipeline
  key: pipelineId
  description: An organization data pipeline. Emits the completion webhook.
- name: UploadPortal
  key: portalId
  description: A branded external upload surface for collecting data from third parties.
  paths: 12
- name: TransformationBlock
  key: transformationBlockId
  description: A custom organization block that transforms data.
- name: AIAction
  key: actionId
  description: A GenAI labelling/transform action with a preview-then-apply lifecycle.
  paths: 5
- name: Whitelabel
  key: whitelabelId
  alt_key: whitelabelIdentifier
  description: A white-label tenancy that rebrands Studio for an organization.
- name: EonTunerTrial
  key: trialId
  description: One trial inside an EON Tuner AutoML run.
- name: PostProcessing
  key: postProcessingId
  description: A post-processing block applied after inference.
relationships:
- from: Organization
  to: Project
  type: has_many
  via: ownerOrganizationId
- from: Organization
  to: User
  type: has_many
  via: organizationId
  note: members
- from: Organization
  to: Bucket
  type: has_many
  via: bucketId
- from: Organization
  to: OrganizationDataset
  type: has_many
  via: dataset
- from: OrganizationDataset
  to: DataItem
  type: has_many
  via: dataId
- from: Organization
  to: Pipeline
  type: has_many
  via: pipelineId
- from: Organization
  to: UploadPortal
  type: has_many
  via: portalId
- from: Organization
  to: TransformationBlock
  type: has_many
  via: transformationBlockId
- from: Organization
  to: Whitelabel
  type: has_one
  via: whitelabelId
- from: Organization
  to: Organization
  type: has_many
  via: innerOrganizationId
  note: White-label parents nest child organizations.
- from: Project
  to: User
  type: belongs_to
  via: ownerUserId
  note: A project is owned by a user or by an organization, not both.
- from: Project
  to: Sample
  type: has_many
  via: sampleId
- from: Project
  to: Impulse
  type: has_many
  via: impulseId
- from: Impulse
  to: DSPBlock
  type: has_many
  via: dspId
- from: Impulse
  to: LearnBlock
  type: has_many
  via: learnId
- from: Impulse
  to: PostProcessing
  type: has_many
  via: postProcessingId
- from: Project
  to: Job
  type: has_many
  via: jobId
- from: Organization
  to: Job
  type: has_many
  via: jobId
  note: Organization jobs are a separate collection with their own status and cancel operations.
- from: Project
  to: Device
  type: has_many
  via: deviceId
- from: Project
  to: Deployment
  type: has_many
  via: deploymentVersion
- from: Project
  to: ProjectVersion
  type: has_many
  via: versionId
- from: Project
  to: ApiKey
  type: has_many
  via: apiKeyId
- from: Project
  to: HmacKey
  type: has_many
  via: hmacId
- from: Project
  to: AIAction
  type: has_many
  via: actionId
- from: Project
  to: EonTunerTrial
  type: has_many
  via: trialId
- from: Sample
  to: DataItem
  type: derived_from
  via: ingestion
  note: Organization data items are imported into projects as samples through pipelines.
notes:
- The graph has no soft-delete or trash entity anywhere. ProjectVersion is the only rollback surface.
- Ids are integers with no type prefix, so an agent must never infer entity type from an id value.
- The Organization subtree is Enterprise-only; on the Developer plan it does not exist at all.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com