I build tools that solve the problems in front of me and make my workflow more efficient.
Current focus: Learning to query data with SQL.
Your agent renders the UI change as ephemeral HTML in your own design system. Annotate your changes, compare the cuts, drag what's misplaced, hit send, and it rebuilds.
The keyboard-first, human-in-the-loop bench for computer vision ML work.
A personal app platform and control plane for local tooling.
The rest of my work lives outside these repos, most of it under my company, EdgeFlow AI.
- The data-pipeline engine that enumerates, filters, and curates public procurement records into verified ground-truth packages.
- I am building a proprietary computer vision model end to end: dataset design, synthetic data generation, training, and evaluation.
- RDF knowledge graphs in JSON-LD with a SKOS taxonomy and my own query engine on top. Plain-English questions compile to graph walks on Oxigraph (SPARQL) and Cayley (Gizmo).
- A model-agnostic agentic harness with a self-hosted control-plane dashboard to run the system.
Open to work in: software and tech-adjacent roles
Email: race.nevils@edgeflowai.com
LinkedIn: linkedin.com/in/race-nevils


