Developer docs / AI Fabric tutorials

Nightly batch job

  • Cron
  • Volume
  • Image DSL

Take it for a spin

The whole program in one file, ready to run.

nightly_batch.py
import bladekit

app = bladekit.app("nightly-batch")

output_vol = bladekit.Volume.from_name("nightly-results", create_if_missing=True)

image = (
    bladekit.Image.debian_slim(python_version="3.12")
    .pip_install("polars", "pyarrow", "httpx")
)

@app.function(
    image=image,
    cpu=8,
    memory=16384,
    timeout=3600,
    volumes={"/results": output_vol},
    schedule=bladekit.Cron("0 3 * * *"),  # 03:00 UTC every day
    retries=1,
)
def nightly_job() -> None:
    import polars as pl
    from datetime import datetime, timezone

    today = datetime.now(timezone.utc).strftime("%Y-%m-%d")
    print(f"Starting nightly batch for {today}")

    # Example: aggregate the day's data (swap this synthetic frame for a
    # real ingestion source, e.g. an httpx.get() call, in production).
    df = pl.DataFrame(
        {
            "region": ["us-east", "us-west", "eu-west", "us-east", "eu-west"],
            "revenue": [120.5, 98.2, 75.0, 40.1, 60.3],
        }
    )
    summary = df.group_by("region").agg(pl.col("revenue").sum())
    summary.write_parquet(f"/results/{today}-summary.parquet")
    output_vol.commit()  # persist writes so other functions can read them
    print(f"Saved {len(summary)} regions to Volume.")

Run it

  1. Deploy: blade deploy nightly_batch.py.
  2. The job runs every night at 03:00 UTC. Go to the Crons page to see the schedule and last run status.
  3. Results are saved to the nightly-results Volume. Mount it in another function to query them.
  4. retries=1 means if tonight's run fails, AI Fabric tries once more before marking it failed.

Concepts used: Deployment, Run

Contact the team

Loading the form…