Browse categories
Explore
Fiverr Pro
English
$
USD
I will build snowflake data pipelines with snowpipe, streams, tasks and dynamic tables
About this Gig
Need data in Snowflake that loads itself, updates only what changed and never double counts? I build pipelines inside Snowflake with plain SQL, so there is no extra tool to pay for or maintain.
What I build:
- Ingestion: stages, file formats and Snowpipe auto-ingest from S3, Azure Blob or Google Cloud Storage
- Change capture: Streams + Tasks with MERGE upserts and de-duplication
- History: SCD Type 2 dimensions for customers, products or accounts
- Reporting layer: Dynamic Tables for daily and monthly marts
- Data quality: checks for nulls, duplicates, orphan keys and freshness
- Monitoring: task history, failure alerts by email and cost guardrails
How I work:
- We agree the sources, tables and refresh times
- I build in a dev schema and test with your sample data
- You review the results and the run history
- Handover with a runbook and diagram
You get version-controlled SQL, a pipeline diagram, test results and documentation.
Already using Fivetran, ADF or SSIS? I can take the data from where they land it.
My Portfolio
Other Data Engineering Services I Offer
FAQ
Why build pipelines in Snowflake instead of another tool?
Snowpipe, Streams, Tasks and Dynamic Tables run inside your account, so there is no extra licence or server. They are a good fit when the sources can drop files or when the data is already in Snowflake.
Can you load from APIs or databases directly?
Yes. For databases and APIs I use your existing tool (ADF, Fivetran, SSIS) or a small scheduled script to land files in a stage, then Snowflake takes it from there.
How fresh will my data be?
Snowpipe usually loads files within minutes of arrival. Tasks can run every minute, and Dynamic Tables follow the target lag you choose, for example 15 minutes or 1 hour.
Will this increase my Snowflake bill?
Tasks only wake the warehouse when a stream has new data, and I use the smallest warehouse that meets the timing. You get an estimate of daily credits before go-live.
What happens if a load fails?
Failed tasks send an email alert, the stream keeps the unprocessed changes, and the next run picks them up. The runbook explains how to check and re-run each step.
