Portfolio · 2026

Most candidates ask you to take their judgment on faith. Mine is published.

Three sites, built and maintained in public. Each one answers a question you would otherwise have to ask in an interview or a scoping call, and lets you check the answer before we ever speak.

01 · The plan

encarnacao.net ↗

“What would you actually do here?”

An interactive simulator showing why faster AI generation rarely means faster delivery, a first-90-days plan with named deliverables, and the four rules I run an AI program by.

• You can evaluate my operating plan before the first conversation.
• I diagnose stalled AI programs as review, QA and release problems, not model problems.
• The first 90 days have a sequence: take inventory, gate one workflow end to end, then make it the default.

02 · The record

bitandbeing.com ↗

“How do you think, and can you explain it?”

More than a hundred dated essays on machine minds, enterprise AI, the human factor, engineering craft, risk and security, and the hardware underneath. Plus a podcast and a newsletter.

• Judgment written in public and dated, so you can argue with it before you hire me.
• Proof I can make technical risk legible to boards, executives and engineers alike.
• A clear voice for your AI story, inside the company and with customers.

03 · The proof

islandsofstability.com ↗The preprint ↗

“Can you prove what you claim?”

A public research notebook: six falsifiable propositions on where complex systems work, pre-registered tests deposited with DOIs, and 18,720 measured runs mapping where AI agents hold up as their tools fail.

• The evaluation discipline your AI program needs before the board asks what it can prove.
• I build instruments, not slides, and publish the unflattering numbers too.
• Comfortable being wrong in public, which is how errors get caught early.

25

Years of engineering leadership

38

Fortune 500 companies advised

50+

Engineers, QA and PMs hired and mentored

11

Countries where I have worked on site