Session

Platform Strategies for AI Models—Cross-Region Migration and DR

Overview

ExperienceIn Person
TypeBreakout
TrackData and AI Governance
IndustryEnterprise Technology, Health and Life Sciences, Financial Services
TechnologiesMLFlow, Mosaic AI, Unity Catalog
Skill LevelIntermediate
Duration40 min

AI for enterprises, particularly in the era of GenAI, requires rapid experimentation and the ability to productionize models and agents quickly. This means that the availability of the entire Lakehouse- from raw data to ETL pipelines to notebooks and ultimately the serving layer- is more important than ever.

 

A strong Disaster Recovery strategy has become a key part of a strong AI practice; as cloud providers struggle with rising demand for GPUs in environments, VM shortages have become commonplace, and add to the pressure of general cloud outages. Enterprises that can quickly leverage GPU capacity in other cloud regions will be better equipped to capitalize on the promise of AI.

 

Leveraging Unity Catalog Models is a strong recommendation going forward for Disaster Recovery readiness and good AI governance practices. In this presentation we will show an end-to-end example of how UC enables DR for fully-governed AI models.

 

Co-presenters: Tony Farias and Greg Wood 

Session Speakers

IMAGE COMING SOON

Greg Wood

/Lead SSA
Databricks

IMAGE COMING SOON

Tony Farias

/Sr Solutions Architect
Databricks