# Generated by API Evangelist (build-phrasing.py). Our phrasing, not observed demand.
overlay: 1.0.0
info:
  title: API Evangelist conversational phrasing for Amazon Neptune Data ML API
  version: 1.0.0
extends: openapi/amazon-neptune-ml-api-openapi.yml
actions:
- target: $.info
  update:
    x-apievangelist-phrasing:
      method: generated
      generated: '2026-10-01'
      generator: build-phrasing.py
      label: Generated by API Evangelist
      operations: 16
- target: $.paths['/ml/dataprocessing'].get
  update:
    x-apievangelist-phrasing:
      intent: List ML data processing jobs
      effect: read
      questions:
      - Which data processing jobs is Neptune ML running for me?
      - Can I cap the number of data processing jobs listed?
      instructions:
      - text: List my Neptune ML data processing jobs.
      - text: Show at most {max} ML data processing jobs.
        slots:
          max: query.maxItems
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/dataprocessing'].post
  update:
    x-apievangelist-phrasing:
      intent: Start an ML data processing job
      effect: write
      questions:
      - How do I prepare graph data in S3 for Neptune ML training?
      - What input and output S3 paths does a data processing job require?
      instructions:
      - text: Start an ML data processing job from {input} to {output}.
        slots:
          input: requestBody.inputDataS3Location
          output: requestBody.processedDataS3Location
      - text: Prepare training data from {input} into {output} for model type {model_type}.
        slots:
          input: requestBody.inputDataS3Location
          output: requestBody.processedDataS3Location
          model_type: requestBody.modelType
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/dataprocessing/{jobId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Check an ML data processing job
      effect: read
      questions:
      - Is my ML data processing job done?
      - What state is a given Neptune ML data processing job in?
      instructions:
      - text: Get the status of ML data processing job {job}.
        slots:
          job: path.jobId
      - text: Check progress on data processing job {job}.
        slots:
          job: path.jobId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/dataprocessing/{jobId}'].delete
  update:
    x-apievangelist-phrasing:
      intent: Cancel an ML data processing job
      effect: destructive
      questions:
      - Can I cancel a data processing job that's taking too long?
      - Will cancelling a processing job clean up its S3 output?
      instructions:
      - text: Cancel ML data processing job {job}.
        slots:
          job: path.jobId
      - text: Cancel data processing job {job} with clean set to {clean}.
        slots:
          job: path.jobId
          clean: query.clean
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltraining'].get
  update:
    x-apievangelist-phrasing:
      intent: List ML model training jobs
      effect: read
      questions:
      - Which model training jobs are running in Neptune ML?
      - What graph neural network training jobs are active?
      instructions:
      - text: List my Neptune ML model training jobs.
      - text: Show at most {max} model training jobs.
        slots:
          max: query.maxItems
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltraining'].post
  update:
    x-apievangelist-phrasing:
      intent: Start an ML model training job
      effect: write
      questions:
      - How do I train a model on my processed graph data?
      - Can I use managed spot training to cut the cost of model training?
      - What does a model training job need from the data processing step?
      instructions:
      - text: Train a model from data processing job {processing_job}, saving to {s3}.
        slots:
          processing_job: requestBody.dataProcessingJobId
          s3: requestBody.trainModelS3Location
      - text: Start training on {instance} from processing job {processing_job} with output {s3}.
        slots:
          instance: requestBody.trainingInstanceType
          processing_job: requestBody.dataProcessingJobId
          s3: requestBody.trainModelS3Location
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltraining/{jobId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Check an ML model training job
      effect: read
      questions:
      - Has my model finished training?
      - What is the status of a specific Neptune ML training job?
      instructions:
      - text: Get the status of model training job {job}.
        slots:
          job: path.jobId
      - text: Check how training job {job} is going.
        slots:
          job: path.jobId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltraining/{jobId}'].delete
  update:
    x-apievangelist-phrasing:
      intent: Cancel an ML model training job
      effect: destructive
      questions:
      - Can I abort a model training job partway through?
      - Is there an option to delete training artifacts when cancelling?
      instructions:
      - text: Cancel model training job {job}.
        slots:
          job: path.jobId
      - text: Cancel training job {job} with clean set to {clean}.
        slots:
          job: path.jobId
          clean: query.clean
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltransform'].get
  update:
    x-apievangelist-phrasing:
      intent: List ML model transform jobs
      effect: read
      questions:
      - Which model transform jobs are active in Neptune ML?
      - What embedding generation jobs are running?
      instructions:
      - text: List my Neptune ML model transform jobs.
      - text: Show at most {max} model transform jobs.
        slots:
          max: query.maxItems
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltransform'].post
  update:
    x-apievangelist-phrasing:
      intent: Start an ML model transform job
      effect: write
      questions:
      - How do I generate embeddings and inference artifacts from a trained model?
      - Where does a model transform job write its output?
      instructions:
      - text: Start a model transform job writing to {s3}.
        slots:
          s3: requestBody.modelTransformOutputS3Location
      - text: Transform training job {training_job} into inference artifacts at {s3}.
        slots:
          training_job: requestBody.mlModelTrainingJobId
          s3: requestBody.modelTransformOutputS3Location
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltransform/{jobId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Check an ML model transform job
      effect: read
      questions:
      - Is my model transform job finished?
      - What state is a given Neptune ML transform job in?
      instructions:
      - text: Get the status of model transform job {job}.
        slots:
          job: path.jobId
      - text: Check progress on transform job {job}.
        slots:
          job: path.jobId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/modeltransform/{jobId}'].delete
  update:
    x-apievangelist-phrasing:
      intent: Cancel an ML model transform job
      effect: destructive
      questions:
      - Can I cancel a model transform job that's still running?
      - Will cancelling a transform job remove its S3 artifacts?
      instructions:
      - text: Cancel model transform job {job}.
        slots:
          job: path.jobId
      - text: Cancel transform job {job} with clean set to {clean}.
        slots:
          job: path.jobId
          clean: query.clean
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/endpoints'].get
  update:
    x-apievangelist-phrasing:
      intent: List ML inference endpoints
      effect: read
      questions:
      - Which Neptune ML inference endpoints are deployed?
      - Can I cap how many inference endpoints are listed?
      instructions:
      - text: List my Neptune ML inference endpoints.
      - text: Show at most {max} inference endpoints.
        slots:
          max: query.maxItems
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/endpoints'].post
  update:
    x-apievangelist-phrasing:
      intent: Create an ML inference endpoint
      effect: write
      questions:
      - How do I serve predictions from a Neptune ML model?
      - Can I choose the instance type for an inference endpoint?
      instructions:
      - text: Create an inference endpoint from training job {training_job}.
        slots:
          training_job: requestBody.mlModelTrainingJobId
      - text: Create endpoint {id} from transform job {transform_job} on {instance}.
        slots:
          id: requestBody.id
          transform_job: requestBody.mlModelTransformJobId
          instance: requestBody.instanceType
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/endpoints/{endpointId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Check an ML inference endpoint
      effect: read
      questions:
      - Is my inference endpoint up and serving?
      - What status is a specific Neptune ML endpoint in?
      instructions:
      - text: Get the status of inference endpoint {endpoint}.
        slots:
          endpoint: path.endpointId
      - text: Check whether endpoint {endpoint} is ready.
        slots:
          endpoint: path.endpointId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/ml/endpoints/{endpointId}'].delete
  update:
    x-apievangelist-phrasing:
      intent: Delete an ML inference endpoint
      effect: destructive
      questions:
      - Can I delete an inference endpoint to stop paying for it?
      - Is there a way to remove the endpoint's S3 artifacts as well?
      instructions:
      - text: Delete inference endpoint {endpoint}.
        slots:
          endpoint: path.endpointId
      - text: Delete endpoint {endpoint} with clean set to {clean}.
        slots:
          endpoint: path.endpointId
          clean: query.clean
      method: generated
      generated: '2026-10-01'