Skip to content

Architecture Diagram

Purpose

A high-level architecture diagram helps visualize how the main building blocks (config, scripts, ETL runners, storage, training, Prefect) fit together.

What You Will Find

  • Layers: Entry points (scripts) → application/orchestration (config load, idempotency, path resolution) → pipeline execution (ETL, training, Optuna, test offline) → storage (local + S3).
  • Data flow: Raw data → TPL/GENKEY → Windows → Features → (optional feature selection) → Training or Optuna → Models and reports.
  • Scheduling: APScheduler (in-process) or Prefect (server + worker) triggering the same pipelines via config.

Where to Look

  • The main Architecture page describes the same structure in text.
  • For ETL-specific design, see ETL architecture.
  • If the repo includes a diagram asset (e.g. SVG or PNG under docs/ or img/), it may be linked from the Architecture page or from this section.

This page anchors the “diagram” concept in the Engineering menu; the canonical description remains the Architecture and ETL architecture docs.