The person behind the page
Neuroimaging scientist. AI evaluator. Neurotech adviser. The NeuroVersatilist is what happens when you spend fifteen years inside brain scanners, pipelines, and models and decide the best thing to do with that is explain it plainly.
A generalist knows a little about a lot. A versatilist goes deep in multiple areas and can move between them, carrying the logic of each domain into the others. The term comes from organizational research on the kind of professional who is most valuable in complex, rapidly changing fields.
Travis fits that description precisely. The neuroimaging work is not background color; it is fifteen years of principal analyst work across three NIH-funded longitudinal cohorts, building and deploying multimodal MRI pipelines covering fMRI, DTI, MRS, and volumetric analysis on HPC infrastructure. The AI work is not hobbyist tinkering; it is subject matter expertise developed across more than ten projects evaluating large model behavior for Outlier AI, with Oracle contributor status. The neurotech work is not consulting-speak; it is active advisory work at MetaBrain Labs, watching the brain-computer interface industry from inside it.
Those three lanes are genuinely distinct. They also talk to each other in ways that most practitioners miss because most practitioners are only in one of them. That cross-talk is what this blog is for.
The humanities background is the fourth element. A creative writing training does not make science softer. It makes explanation more precise. The goal is never to dumb things down; it is to remove the unnecessary friction between what the science actually says and what a careful, motivated reader can understand.
Active advisory role in the brain-computer interface space. Tracking where the technology actually is, not where the press releases say it is.
Oracle contributor status across 10+ projects evaluating large language model behavior in neuroscience and biomedical domains. The work is essentially: find where the models are wrong, and explain why in terms the models' developers can act on.
Led multimodal MRI pipeline development and deployment for three longitudinal cohorts on HPC infrastructure. Modalities: fMRI, DTI, MRS, volumetric. Output: 11 peer-reviewed publications, h-index 10, 370+ citations. The work was not just analysis; it was building the infrastructure that made the analysis possible at scale.
The scientific training. Also: an undergraduate humanities and creative writing background that never entirely went away, and probably explains why the explanations tend to work.
Neuroimaging methodology, network dynamics, reproducibility, and what the published literature actually establishes versus what the headlines say it establishes. Written by someone who has spent a decade scrutinizing these papers for a living.
What machine learning in neuroscience looks like from inside a real pipeline. Model evaluation, failure modes, capability claims that hold up and ones that don't. Not a cheerleader's account and not a doomer's account.
Where the brain-computer interface field is headed versus where it is being sold as going. The view from inside an active advisory role, not from press releases and conference abstracts.
No cadence for the sake of cadence. New essays and notes go out when they are ready and worth your time. Free to start.
You're on the list. Talk soon.
Free forever, for now. Paid tier coming later.