Between my second and third year of engineering, I spent the summer interning at a data science consultancy.
I primarily worked on projects for Watercare, the company which manages Auckland’s drinking water and wastewater infrastructure.
My research was primarily focused on wastewater overflow prediction, which involved physically modelling wastewater stations across Auckland, and working with sensor data to predict the risk of an overflow. The intention was for this to be surfaced to administrators so that they could prioritise where staff were sent based on the data available. I pushed for the perspective to the project lead that, while prediction was possible, they needed to invest getting consistent data across their infrastructure in order to make the predictions useful enough for decision making. To back this up I walked them through an analysis I performed on the data we were getting back from station’s sensors, which highlighted data integrity issues we were seeing at the source.
Toward the end of my time there, I worked on automating asset renewal analytics for wastewater infrastructure using robotics and custom convolutional neural networks. I wrote the initial architecture for the project and conducted initial experiments before returning to my studies.