Full-stack engineer, fifteen years across automotive, finance, defense, healthcare, and agriculture: General Motors, JPMorgan Chase, L3Harris, T. Rowe Price, Nutrien Ag Solutions, RumbleOn, Integrated Auction Solutions, Brierley + Partners, Kofile Technologies.

Now Lead AI Engineer at OffensiveEdge, building an end-to-end machine learning platform and the LLM product on top of it: training and serving, a ReAct agent harness with tool-calling and hybrid retrieval, and a public prediction ledger anchored to Bitcoin via OpenTimestamps and to the Sigstore Rekor transparency log. Earlier, lead developer on the first multilingual EMS communication platform: Texas Fire Chief's Lone Star Achievement Award, EMS World Top Innovation Award, a patent, acquired by ESO Solutions.

Nearly all of that is private. This account is where the open source is, and it started as an experiment: I never had time to publish, so could an LLM get me to a bar I'd sign my name to without taking anything from primary work? The code was built with an LLM. I'm not hiding that, and I'm not selling it either.

The bar is mechanical. Coverage floors, accessibility gates, payload budgets, and visual baselines run in CI, and generated code clears the same standard as hand-written or it doesn't merge. Axe runs on every page, in both themes, because I spent years on ADA / WCAG / Section 508 remediation at Fortune 500 scale. Users don't grade your code, they grade whether the thing works, so the scoreboard is downloads.

nolindnaidoo.com · LinkedIn is the fastest way to reach me.

letools.dev — 16 tools, all 16 published to Open VSX and the VS Code Marketplace, and all 16 on the official Model Context Protocol registry so agents get the same interface editors do. Counts come from the registries' own APIs, not from me.

Local, deterministic, exit codes as the API.