Combat engagement outcome simulator

Mechanistic interpretability is the reverse engineering of black box—really opaque—deep learning models to understand how they work (think artificial neuroscience). Here it is performed on prime number classifying attention models.

Uiua based swarm intelligence project

An ongoing data scientific exploration of activity in human vusual cortex.

We work with designer Frederik Liederley on a variety of projects, including the development of the Liderley website. His work has been recognized by titans of the fashion industry, including Rick Owens, Raf Simons, and Ann Demeulemeester.

In our collaboration with Potemkin, we provided technological assistance that seamlessly integrated into their exhibitions, creating an effective platform to showcase their captivating narratives and experiences.

Infering the change in culture over time from a corpus of text is a difficult task. This project is an ongoing attempt to do so using deep learning. Creating a high quality mapping between content and culture is a prerequisite for the regulation and improvement of social media platforms.