Avery Yen

Hi, I’m Avery. I’m a first year MS candidate in Data Analytics Engineering at Northeastern university, and a Research Assistant to Professor David Bau at the National Deep Inference Fabric, where I study mechanistic interpretability of large language models.

Before that, I spent over 12 years as a software engineer, with my longest tenure at Pivotal Labs.

Research Interests

tl;dr What can AI do for us, and what are the right things to do with AI, and can we intersect those two things?

I believe that understanding how AI systems form internal representations and make decisions is essential for building technology we can trust, especially as we continue to hand them more ways to affect our daily life.

My work focuses on developing empirical methods to reverse-engineer the computational mechanisms within neural networks, building tools that make these massive systems transparent and steerable.

I’m particularly interested in interpretability techniques that support beneficial applications in human society: healthcare, sustainability, and economics. If we can recover insights from AI systems, we stand to gain a lot of insights into real problems facing the world today.

Selected Projects

• National Deep Inference Fabric: The NSF National Deep Inference Fabric (NDIF) is a research computing project that enables researchers and students to perform mechanistic interpretability research on models, with sizes up to a 405B parameter open-weight model.
• Fine Tuning BERT on Trump Socials & VOO: My in progress work showcases one area where I aim to apply mechanistic interpretability to recover previously-unknown understanding into a complex socio-economic system.

Contact

Feel free to find me on LinkedIn.