3+ hour, 26+ min ago  (832+ words) Our results demonstrated up to 2. 5x faster distributed Py Torch training (use case 9 in TPCx-AI) and up to 1. 8x faster distributed XGBoost training (use case 8 in TPCx-AI), with up to 8" lower per-run infrastructure costs in those tested environments compared to Databricks. The…...

1+ day, 4+ hour ago  (590+ words) You can make AI_EXTRACT even more accurate by fine-tuning the underlying Arctic-Extract model and realize the following benefits: Arctic-Extract is a proprietary, vision-based large language model purpose-built for high-quality key information extraction across unstructured data, ranging from documents, images, to raw…...

1+ day, 22+ hour ago  (711+ words) Generalized Skew Handling (GSH) is a Snowflake query execution feature that automatically detects data skew at runtime and redistributes work to idle threads " without any user configuration. GSH processes tens of millions of queries daily, delivering an average 3. 46% execution time…...

1+ week, 3+ day ago  (701+ words) If you want the broader architectural picture, start with the intro post on the Snowflake Model Serving. This post focuses specifically on async batch inference. Please refer to this article for comparison of various offerings in Snowflake Model Serving. You…...

2+ week, 1+ day ago  (486+ words) For most of my career in FP&A, the work started after the numbers were closed. We would gather data, reconcile it, explain it, package it and then deliver it to the business. The problem was that by the time…...

2+ week, 2+ day ago  (540+ words) Cortex Code security skills integrate these journeys into a single, secure, AI-assisted workflow. Our goal is to make complex security administration feel less like struggling with a configuration matrix and more like a guided dialog with a specialized teammate. The…...

2+ week, 6+ day ago  (773+ words) Bad data is expensive. Not because it's hard to find, but because you find it too late. An order lands with status = 'actve' instead of 'active'. A JSON event arrives with an amount: "banana". A NULL slips into a nonnullable…...

3+ week, 1+ day ago  (731+ words) Large language models are rapidly evolving into autonomous agents capable of executing complex, multistep workflows. One of the most important frontiers in this evolution is the data science agent " an AI system that can autonomously ingest data, engineer features, train…...

3+ week, 3+ day ago  (649+ words) Data integration is one of the first and most critical steps in building any data pipeline. It's how raw data becomes usable, trusted, and ready for downstream applications. But in practice, it's also where teams lose the most time. Connectiving…...

3+ week, 4+ day ago  (541+ words) Interactive warehouses are purpose-built for speed. They are optimized for short, highly concurrent reads, the kind that power dashboards, data APIs and AI agents querying structured data in real time. We're introducing Fallback Warehouse support for interactive warehouses. With a…...