Pipeline Orchestrator

A real-time visualization of a data pipeline DAG — the kind of execution graph you’d see in Airflow, Dagster, or Prefect. Nodes progress through stages (ingest, validate, enrich, transform, join, load), with fan-out parallelism, dependency resolution, and automatic failure recovery.

How it works

The orchestrator resolves a dependency graph at each tick. A node transitions from queued → running → done only after all upstream dependencies complete. Data packets animate along cubic Bezier edges to visualize throughput between stages.

Failures are injected randomly — a node turns red, pauses, then retries automatically. The sparkline in the footer tracks pipeline completion runs over time.

Architecture

Built entirely with Svelte 5 runes and SVG. No external animation or charting libraries. The DAG layout is hand-positioned for clarity, with spatial relationships reflecting the actual data flow topology.