01. Problem
A San Francisco startup building a custom machine learning system fed by data stored in Salesforce invited us to build a custom integration to regularly move high volume Salesforce data to Amazon AWS to populate machine learning jobs.
02. Solution
A high-performance MuleSoft integration was created to identify any new data in Salesforce apply required transformations and incrementally move any new chunks to Amazon AWS S3 system being a load area for Amazon Elastic MapReduce service. The integration was deployed in the cloud with an option to scale up and down depending on the actual input volumes.
03. Results
Custom integration built in MuleSoft enabled us to create a fast data loading solution that identifies, transforms and sends to AWS only the newly updates entries in Salesforce. The process incrementally populates the machine learning algorithms on regular basis keeping the artificial intelligence model and resulting reports always up-to-date. It also limits the required resources and cloud costs as only incremental updates. not full snapshots, are executed.