stage 01 / commitpassed
Adrian Dantas
commit 7f3c9ad · HEAD → main
author  Adrian Dantas <adrian@adriandantas.com>
target  production

Adrian Dantas

Product engineer who ships. Now pioneering AI in DevOps to build what’s next — adopting what measurably moves delivery, and dropping what doesn’t.

stage 02 / buildpassed

26 years of shipping real products.

Before it was AI, it was product. I’ve spent my career turning ideas into software that ships and holds up in production — designing systems, leading engineers, and owning the path from first commit to what real users depend on.

Deep technical work paired with product judgment: what to build, what to cut, and how to make delivery boring in the best way — automated, self-serve, and reliable enough that teams stop thinking about it.

stage 03 / verify2 promoted · 2 rejected

I keep what proves out. I cut what doesn’t.

New tech earns its place by evidence, not by being new. This is how I run the frontier — not AI for AI’s sake.

  • Give agents a real harness, not guesswork

    MCPs that feed coding agents compliant CI/CD pipelines and searchable platform & SDLC knowledge — so they build on the org’s real conventions, not plausible guesses. promoted

  • Make the design system AI-ready the same way

    An MCP that hands agents proven, accessible components instead of generating UI from scratch. verified

  • Chasing every new model on its leaderboard

    I run Claude Code, opencode, and Pi on real work and keep the setup that ships — not the one that tops a benchmark. rejected

  • Swapping the stack for hype

    I test frontier models hands-on, including open ones — most “SOTA” gains vanish outside the benchmark. rejected

— verified in production at

stage 04 / shipin progress

Now pioneering AI in DevOps — to build what’s next.

Staff Engineer · contract

Upwork

Where the day-to-day pioneering happens: bringing AI into DevOps and developer workflows — the parts that measurably move delivery, not the demos.

Co-founder

PrimeDX

My studio for launching products — the vehicle for turning what proves out on the frontier into things people can actually use.

stage 05 / deployarmed

Let’s ship something.

Building a product, or bringing AI into how your teams deliver? Start a conversation — no pitch, usually a quick reply.

or reach me on