About

Skillset | Projects | Results

I’m Daniel Lumian. I run Dexterous Data as a solo consultancy focused on data science methodology, ML validation, and helping technical teams work more effectively with their data.

Background

My training is in research — a PhD in psychology with 15+ years of experience designing experiments, analyzing complex datasets, and building analysis pipelines in neuroimaging and fMRI. I’ve worked in academia, industry (including a Big 4 firm), and on government research contracts with the NIH and FDA.

That research background shapes how I approach consulting work. I think about whether the method fits the question before thinking about whether the code runs. I care about validation, reproducibility, and whether results will hold up outside the conditions they were tested in.

Open Work

The best way to evaluate whether I’m a fit for your problem is to look at how I work:

Pixel Process is my open-source platform — applied project case studies, workflow patterns, and interactive data science materials. Everything is free, and the full source code is public.

Face Value is my current flagship project — an emotion recognition pipeline demonstrating that domain-matched weak supervision outperforms benchmark-trained models on ecological validation tasks.

These projects show how I think about problems, make decisions, and document tradeoffs. They’re a more honest representation of my work than any list of skills or tools.

Get In Touch

Find me on GitHub.