Ex-cybersecurity engineer, currently a research engineer studying and using mechanistic interpretability style tools and techniques to extract learned structure from different types of Neural Networks. I am also very interested in pre-training and creating better evals.