slug: tensorwave
provider: TensorWave
generated_by: planning/capability-mapping/scripts/classify_capabilities.py
model: claude-opus-5
frame:
- Electrical Components & Equipment
min_confidence: 0.7
capability_model:
  source: https://github.com/vincentmakes/turbo-ea-capabilities
  license: CC-BY-4.0
  attribution: Turbo EA Capabilities by Vincent Verdet — Turbo EA, https://github.com/vincentmakes/turbo-ea-capabilities,
    CC BY 4.0
  notice: NOTICE
edge_count: 3
edges:
- tag: Training
  spec_file: tensorwave-training-api-openapi.yml
  capability_id: BC-610.60
  capability_id_l1: BC-610
  capability_name: Artificial Intelligence Management
  confidence: 0.85
  evidence: POST /v1/megatron/train 'Submit a training job'; GET /v1/megatron/train/{job_hash}/checkpoints
    'List checkpoints written by a training job'; POST .../publish 'Publish a checkpoint to Hugging Face
    Hub'
  reason: Model training job submission, training dataset upload, checkpoint inventory and model publishing
    constitute the ML model lifecycle / MLOps, which is Artificial Intelligence Management.
- tag: Generate
  spec_file: tensorwave-generate-api-openapi.yml
  capability_id: BC-610.60
  capability_id_l1: BC-610
  capability_name: Artificial Intelligence Management
  confidence: 0.75
  evidence: POST /v1/generate 'Submit a batch of prompts to the inference queue'; GET /v1/generate/metrics
    'Inference queue metrics'
  reason: The operations run LLM inference workloads (batch prompt submission, queue work distribution,
    adaptor listing) on the vendor's training/inference stack. That is AI/ML model serving and MLOps,
    i.e. Artificial Intelligence Management. Some ambiguity because much of the surface is queue plumbing
    rather than model governance, hence 0.75.
- tag: OpenAI
  spec_file: tensorwave-openai-api-openapi.yml
  capability_id: BC-610.60
  capability_id_l1: BC-610
  capability_name: Artificial Intelligence Management
  confidence: 0.75
  evidence: POST /v1/chat/completions 'Create a chat completion'; GET /v1/models 'List available models'
  reason: An OpenAI-compatible inference surface exposing model listing and completion generation — operating
    ML models in production, which falls under Artificial Intelligence Management. It is a thin surface,
    so 0.75 rather than higher.