Zach Maas

• Independent mechanistic interpretability researcher ⋅ PhD in ML for computational biology

Hi! I’m Zach. I’m currently an independent researcher studying mechanistic interpretability on a Coefficient Giving TAIS grant. My work focuses on model interpretability using conditional dependence as a lens for understanding structure in sparse bases. During my PhD, I worked on trying to understand interpretability in biological models where we don’t have the same semantic grounding that’s present in language.

Boulder, CO ⋅ hey@zachmaas.com ⋅ @zmaas ⋅ zach-maas

News

• 2026/07: Presented spotlight in the Mechanistic Interpretability Workshop at ICML 2026
• 2026/01: Started Coefficient Giving TAIS grant
• 2025/05: Completed my PhD!

Highlighted Work:

• Conditional Dependence Structure in Sparse Autoencoder Features: ICML Mech Interp Workshop 2026 spotlight. Dependence graph structure in SAEs
• Graph Structure Compresses Contrastive Neuron Steering Targets: Can we steer LLM behavior using only MLPs + structured dependence?
• Preliminary Results on Graph SAEs: Building dependence graphs off of SAEs to understand their structure
• Research Notes: Sequence to Read Models: Notes on how to build (and not build) sequence-to-read genomic models

Papers I Liked Recently:

• Subliminal Learning Is Steering Vector Distillation
• Language Model Circuits Are Sparse in the Neuron Basis
• Do Sparse Autoencoders Capture Concept Manifolds?
• Targeted Neuron Modulation via Contrastive Pair Search