Databricks provides built-in AI functions that can be used directly in SQL or notebooks, without managing models or infrastructure.Example:SELECT ticket_id, ai_generate( 'Summarize this support ticket:\n{{text}}', 'databricks-dbrx-instruct', descript...
With AI/BI Dashboards, a best practice is for the creator/owner to 'Schedule' the Dashboard to rerun the underlying datasets when changes have occurred. This ensures the Visualizations are rendered with the freshest data. But users still will questi...
@mark_ott That is a nice idea and quite useful. Quick question: how did you define 'stale' data in your case? So what is the threshold at which your 'conditional equation' color codes the date red? did you somehow link that to the refresh schedule?
One thing becomes very clear when you spend time in the Databricks community: AI is no longer an experiment. It is already part of how real teams build, ship, and operate data systems at scale.For a long time, many organizations treated data engineer...
When Databricks One was launched, the default behaviour of the system-managed users group was a major issue. Since every new user is automatically added to users group, and that group traditionally came with "Workspace Access" entitlements, admins ha...
Here is a very short video on how to use collect() method with caution as it could overwhelm the cluster in some big data scenarios!!!! Now only in Spanish but I guess youtube will do the trick to translate into very correct English xDD[video]
The Industrial Data ChallengeManufacturing enterprises today operate with a fundamental paradox: they're drowning in data yet starving for insights. A typical plant generates terabytes of information daily across dozens of systems—from shop floor PLC...
In this article, I’ll walk you through transforming a basic PySpark notebook into a production-ready data pipeline with comprehensive quality checks using Great Expectations patterns in Microsoft Fabric and Databricks. We’ll start simple and progress...
It is finally possible to deploy dashboards using DABS and change the catalog and schema. It is solving the biggest problem with bringing the dashboard to production. New parameters for the dashboard resource were added: dataset_catalog and dataset...
@Hubert-Dudek - This was really a much-awaited feature. Eventually DAB must cover all developed assets which requires promotion to higher environment to cover the CICD seamlessly.
We've all been there. You need to loop through a list of tables, apply some transformation, handle a few edge cases, and maybe catch an error or two. In most data platforms, that means switching to Python or Scala—even if 90% of your logic is just SQ...
Correction: Preview is available on Databricks Runtime 16.3 and above.
As data platforms mature, the focus is no longer just scalability—it is about speed, simplicity, and cost efficiency. Engineering teams want to deliver insights faster without managing infrastructure, while organizations want predictable costs and st...
The latest update for the first week of 2026 is the addition of window functions in Metrics View. In enterprises, there are always measures like cumulative sales or rolling forecast, so it is really important that we can use window functions in busin...
@Hubert-Dudek - Thanks for sharing these bite-size updates as these get lost in many release-notes. Keep it up.
Assigning the result of a shell command directly to a Python variable. It is my most significant finding in magic commands and my favourite one so far. Read about 12 magic commands in my blogs: - https://www.sunnydata.ai/blog/databricks-hidden-magic-...
How Digital Payment Lending Platforms Can Collaborate with Banks Without Exposing Sensitive Data1. Business Context & Regulatory RealityIn 2020, large Indian fintech platforms faced a unique regulatory constraint: NBFC‑led digital platforms were not ...
%%capture magic command not only suppresses cell output but also assigns it to a variable. You can later print cell output just by using the standard print() function #databricksRead about 12 magic commands in my blogs:- https://www.sunnydata.ai/blog...
Last from "everywhere" improvements in Spark 4.1 / Runtime 18 is IDENTIFIER(). Lack of support for IDENTIFIER() in many places is a major pain, especially when creating things like Materialized Views or Dashboard Queries. Of course, we need to wait a...
Thanks for sharing @Hubert-Dudek !
| User | Count |
|---|---|
| 85 | |
| 75 | |
| 72 | |
| 60 | |
| 43 |
