Databricks Clean Rooms Product Demo

my name is nille Gad I lead product management for data Andi collaboration at data brecks today I'm excited to walk you through a demo of data R clean rooms as a quick refresher clean rooms enables organizations to securely collaborate with their customers and partners on sensitive data without compromising privacy for this demo imagine that I'm a member of a fraud detection team at a large Bank bank a and we're collaborating with another financial institution Bank B to conduct a joint fraud analysis I as Bank a have data about my customers their transactions history and also a proprietary AI model that have built that analyzes transactions data my collaborator Bank B also has data about transactions merchants and customer risk profiles that they want to use however there are a few challenges we need to overcome first we cannot share sensitive information about our customers or transactions with each other second our data is on different clouds and data platforms but we still want to be able to collaborate without having to replicate or move the data finally we also want to leverage proprietary code and AI models inside of a clean room for analysis we're going to use datab brick clean terms to overcome these challenges and collaborate in a secure privacy safe environment to identify fraudulent activities let's check it out all right I'm going to start off in the workspace of Bank a and in a few steps I will go create a cleaner I'll give this cleaner a name and for my cloud region I'm using it us and US West 2 what's amazing is that it doesn't matter whether my collaborator is on different Cloud regions or data platforms they can still participate in the cleaner I know that my collaborator Bank B is already on data bricks so I will invite them using their clean room sharing identifier lastly I can also choose to restrict all outbound access to the internet and network resources but I'll keep this one simple and proceed with full internet access once the cleaner is set up I can now start to bring in my data assets first I will add my transactions and customers data I can also add unstructured data and AI models so here I'm going to include a private library that calls an AI function to help me with my fraud detection task my clean room is ready for my collaborator to join let's switch gears now I'm Bank B and I can join the clean room that has been created by my banking partner and see all the pre- added data assets and actually when I click into the transactions data table you can notice that I can only see the underlying metadata but not the actual data itself now it's my turn to bring data into the cleaner so as Bank B I'm going to add in my customer risk profiles and all the fraud transactions data which is actually sitting in the snowflake data warehouse I can use lake house Federation to connect to my snake Warehouse without needing to do any ETL or data copy next I choose to add a notebook that I can collaborate with that is mutually agreed between me Bank B and Bank a this notebook essentially will let us analyze both of your data assets and identify fraudulent transactions let's move ahead to bank a once I hit refresh I can see that there is a new notebook that has been added into the cleaner I can inspect the notebook making sure it contains all the code that we mutually agreed on and choose to run it this will then kick off a serverless job and within a few seconds I should be able to see the results of my analyses all right I can see the output generated by The Notebook which even contain some rich visuals that help me understand which transactions I can flag based on things originating country all right let's summarize what we just saw two organizations were able to bring their respective data assets into a secure data RIS clean roomm without needing to expose or reveal the underlying data they were even able to bring their proprietary ml or AI models it didn't really matter if they were on different clouds regions or data platforms they were able to use the power of Delta sharing and lakeh house Federation to share data assets lastly they are able to run mutually approved jobs against their data to detect fraud patterns I hope you enjoyed this demo thank you