Kyle Chakos

Worked with Shell International Exploration & Production and other seniors to improve their drilling techniques. We devised a machine learning model that analyzed data in real time and compared it with previous drilling data; we identified the type of rock being drilled through and provided recommendations on how to dig through the rock faster and safer.

We accomplished this by utilizing a mixture of the Gaussian model to identify the rock and statistically comparing the identified cluster to other similar clusters to provide better recommendations. All of the code for this project was written in Python.

This machine learning model was used to detect fraudulent credit card usage within 0.5 seconds of the request for purchase. In addition, we added A/B testing for the model deployment, enabling its continuous improvement.

I was in charge of setting up and architecting the back end of this service, which primarily utilized Kafka and Java to ensure everything ran quickly. Our machine-learning models were deployed using Amazon SageMaker.

A process to quickly undertake all CCPA deletions and access requests within 30 days of receiving the request. The process was previously hosted on a Google Sheet and required three people to complete. We utilized OneTrust to collect and track requests and integrated them with the API. This enabled us to successfully delete all user data and retrieve it when requested without human intervention.