I build systems that process complex data, and I'm fascinated by how they work. These days, I'm deep into understanding how machine learning systems process information - particularly how transformer models handle different types of data, and how we might make their decision-making more transparent.I've spent my career building data systems that matter - from preserving digital history at the Internet Archive to improving public records request infrastructure at NextRequest. Now I'm applying that systems-level thinking to machine learning, trying to peek inside the black box and understand how these architectures really work.Recently, I've been exploring some big questions: How does information actually flow through transformer layers? Could we process audio more effectively by treating it as a sequence rather than a 2D spectrogram? When an ML model classifies a sound, which parts of the audio were most important to its decision?I'm particularly interested in projects that use ML to address environmental and scientific challenges. Currently working on processing underwater microphone recordings to identify and understand marine mammal vocalizations.
Listed skills include Web Applications, Javascript, Perl, Databases, and 30 others.