This series investigates how omnipresent machine learning algorithms see our faces under the hood of the algorithms. I generate elemental ghostly faces known as eigen faces which capture the mathematical essence of multidimensional training data. When layered together they can create a new face, or they can be used to identify a specific person. The digital eigen faces outputs are integrated into physical artworks using various experimental layering techniques along with paint, mixed media and found objects. The artworks shown here are all generated from a dataset of ten years of selfies and family photos. The portraits peer inward and also reflect the wider context we all live in today with this omnipresent face technology. Contact me if you are interested in a commisison.