Skip to content

Getting started with built-in nodes

This guide shows how to build a small pipeline using built-in nodes from meridian.nodes.

Install

Bash
uv pip install meridian-runtime

Create a simple pipeline

Python
from meridian.core import Scheduler, SchedulerConfig, Subgraph
from meridian.nodes import DataProducer, MapTransformer, DataConsumer

p = DataProducer("p", data_source=lambda: iter(range(10)), interval_ms=0)
m = MapTransformer("m", transform_fn=lambda x: x * 2)
seen: list[int] = []
c = DataConsumer("c", handler=seen.append)

g = Subgraph.from_nodes("g", [p, m, c])
g.connect(("p", "output"), ("m", "input"))
g.connect(("m", "output"), ("c", "input"))

s = Scheduler(SchedulerConfig())
s.register(g)
import threading, time
th = threading.Thread(target=s.run, daemon=True)
th.start(); time.sleep(0.2); s.shutdown(); th.join()
print(seen)

Next steps

  • Explore categories in the reference for built-in nodes.
  • Add reliability with ThrottleNode, RetryNode, or CircuitBreakerNode.
  • Use NodeTestHarness to test nodes in isolation.