Script to Automate Manual Entry of Database Records• Developed python script to automate navigation through UI and update records necessary for configurations to reduce manual efforts• 2,000 records were regularly required to be updated per week manually spread across 4 FTEs. This resulted in 7-8 hours per week being spent on manually updating records• Reduced time to 30-45 minutes per week only requiring 1 FTE to run and periodically monitor the results of the scriptReporting Data on Customer Journey for Adding/Replacing a Vehicle• Developed data mining script to report on customer activity from phone channel to online channel for adding or replacing a vehicle on their policy• Stitch data from multiple sources (phone and online) to find meaning from data• Collected metrics such as number of: customers calling back, customers going online, etc. and packaged data into excel format to be able to present to business partners Repurpose Dynamic Routing Solution for Phone Transfers• Modified existing solution to dynamically route contacts to representatives for transfers• Leveraged AWS solutions such as Lambda and DynamoDB to overcome existing issue of the original solution only accounting for one queue when representatives had up to 50 queues for the states they were licensed in• Original set of representatives that were the only one receiving transfers had transfers of upwards 60,000 per month. This was reduced to 42,000 after properly distributing transfers to other groups of representatives

- Researched methods to improve security in federated machine learning- Researched the similarities between checkpoints in training phase of deep learning to reduce total size of stored checkpoints and improve efficiency of checkpointing- Modified workflow of Amazon Sagemaker to write checkpoints to persistent memory compared to traditional disk methods