published notes

Description

Had an oportunity to be a part of Data Science team of an industry leading, vertically-integrated digital financial services platform and lender facilitating a broad range of residential real estate loans for professional investors. Providing an impactful solutions to speed up and automatize critical solutions.

My contributions:

  • Developed a machine-learning-based security model masking sensitive information such as SSN and Criminal History in messages between TPOs and the Sales Department, scaling it to 20,000 messages per minute.

  • Designed a model storage system that allows an easier switch and efficient check between versions of security models stored on the Google Cloud.

  • Contributed to analyzing response time and conducting sentiment analysis on messages, leading to the development of a notification system for enhanced TPO-Sales Department communication.

  • Improved the performance of the security models by the method of multithreading on the remote virtual machines using Google Cloud services by 75 percent. (Going from 20 minutes per 100,000 messages to 5 minutes)