Engine Services

The Engine Services layer provides AI/ML pipeline capabilities:

• Pipeline Service - Manages Haystack pipeline definitions, validation, and execution (sync/async)

• Training Service - Trains binary SVM classifiers using Haystack training pipelines

• Prompt Service - CRUD operations for LLM prompt templates stored in the database

• Token Usage Service - Extracts and tracks LLM token usage from pipeline executions

All AI/ML workflows are implemented as Haystack pipelines. The PipelineService is the central orchestrator that builds and executes pipelines from YAML definitions, while delegating component validation to the ComponentRegistry and document validation to the DocumentContract.

Engine Services

• Pipeline Service
  • Key Dependencies
  • Execution Flow
• Training Service
• Prompt Service
• Token Usage Service
  • Key Functions