Zach Maas is an independent mechanistic interpretability researcher based in Boulder with nine years of experience bridging interpretable ML and domain science. He focuses on improving sparse autoencoder (SAE)-based tools and graphical methods to extract circuit-like structures from large language models. Zach completed a PhD applying interpretable ML, transformers, and MCMC normalization techniques to genomic and transcriptional sequencing problems, bringing rigorous experimental design to model introspection. His undergraduate training in chemistry and mathematics informs a quantitative, systems-first approach to mechanistic questions. Now freelancing, he translates research-grade interpretability methods into practical analysis workflows for LLMs. He combines deep domain expertise from genomics with active contributions to ML interpretability techniques that emphasize visual, circuit-centered explanations.
9 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Colorado Boulder
Port of evil-unimpaired code from spacemacs, for general evil use.
Contributions:11 commits, 3 PRs, 7 pushes in 2 years 3 months
unimpairedspacemacs
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.