I worked in the Materials Innovation Factory (MIF), a world-leading research facility, with Unilever on their dedicated floor in the building, in the team responsible for automation and standardisation of the laboratory robots, equipment and data pipelines. Through the Unilever’s in house data entry systems, I worked processing the data that the laboratory equipment produces. Developing my Python skills as I wrote scripts to process outputs from the robots used across MIF, to simplify the regular generation of visualisations that provide at a glance, up to date details of critical metrics. Through communication with colleagues with a chemistry background, to understand what goes into making Unilever products, I was able to extract the questions that they would like visualisations of the data to help them answer. I worked with Power BI to develop a variety of models and visualisation of Unilever and Competitor products to provide insight into the direction for further research and development.

Responsible for working with managing updates and replacements of next-generation technologies on computers running essential laboratory and automation equipment throughout the Materials Innovation Factory (MIF). Worked with colleagues across the MIF to make sure vital data was not lost and that downtime, while improvements were made, was limited to the bare minimum.

I spent my summer studying Deep Learning with the guidance of my research project supervisor. I built, tested and trained Machine Learning models, gaining an understanding of Deep Learning, Neural Networks theory and fluency in Python. My research was focused on Graph Neural Networks, a novel form of Artificial Intelligence that focuses on developing and understanding algorithms that will work with Non-Euclidian data such as social networks, sensor networks and point clouds. I implemented cutting edge models by reviewing leading current academic research on GNNs required to work with Non-Euclidian data. I gained familiarity with Pandas and other python data science packages, as well as a deep understanding of PyTorch as I worked to understand the implementation of the PyTorch Geometric extension library. I continued this project in my final year of university, working specifically point clouds that have the potential to provide improved performance over current standard Euclidian data-based computer vison. I have also built a working knowledge Linux systems using bash and Slurm Workload Manager on the universities High-Performance Computing cluster.