How Cassandra handles users’ data to deliver personalized Marketing Mix Models
In this page you will find:
First of all it is important to know that Cassandra does not gather any user-level data.All of the data we fetch, store and compute is aggregated
Step #1 – Fetching Data
Our current ETL Provider (that handles connections to the users’ platforms and download of the data) is Dataddo. Dataddo is compliant with the Highest Security Standards including SOC2 Type II and GDPR, full documentation can be found here: link to Dataddo’s Security Documentation.
Step #2 – Centralizing & Aggregating the Data
All of the data we download from Dataddo get centralized in Google BigQuery.
The data is queried and modified by Cassandra through Google’s Official Python APIs.
Step #3 – Modeling your Data
Once a Model Training is started the data get uploaded to a dedicated Virtual Machine instance in Scaleway. Being one of the main European’s Cloud Provider, like Dataddo, their are certified upon the main standards: full Documentation here.
The data is uploaded to the Virtual Machines through encrypted SSH Tunnels between Cassandra’s Server & the Virtual Machine itself.
As soon as the training completes, all the data and outputs are downloaded to Cassandra’s servers and removed from the Virtual Machine which is then turned off.
Step #4 – Cassandra’s Platform
The platform itself is deployed on AWS, following the AWS Well Architected Framework, including VPCs, Load Balancers, Encryption of the Data and such.
