Your AI works in the demo.

Most AI problems are decision problems: what the system should decide, what it shouldn’t, and how you’ll know it was right. I work on them as an architect, a product lead and an engineer, usually all three.

Start a projectFive engagements, two weeks to six months.

Hire me full timeHead of product or engineering, head of AI, or staff and principal roles.

Engagements

I take on a small number at a time.

  1. $6MAudited part of a $6M PwC engagement, then helped rebuild the natural-language BI it covered.

    What you get, for example

    Keep

    • Retrieval over your docs
    • The model for open-ended questions

    Make deterministic

    • Dates and time ranges
    • Who’s allowed to see what
    • Which query runs

    Measure

    • Accuracy by question type
    • Cost per answer
    1. 1 Move date math out of the prompt
    2. 2 Put permissions in front of the model
    3. 3 Trim the schema it sees

    AI architecture review

    Your AI works in the demo and wobbles in production.

    A written map of what to keep, what to make deterministic and what to measure, with a ranked plan.

    Two weeks

  2. 3inventions inside GARDE1, a company I built end to end. Two are patent pending.

    An AI product, from zero

    You have the idea and the market, and no one to build it.

    Thesis, design, brand and a production system your first hires can own.

    Eight to sixteen weeks

  3. 99.9%uptime at Pangea (YC W21), from several crashes a day.

    Turnaround and fractional CTO

    The platform is crashing, slowing or outgrowing its first architecture.

    A stable platform, standards that stick, and a team that runs it without me.

    Three to six months, part time

  4. SOC 2led at Litehouse; CMMC is GARDE1’s whole business.

    Enterprise and compliance readiness

    The security review has become the sales cycle.

    SOC 2 or CMMC groundwork, access control, and hardened secrets and pipelines.

    Four to eight weeks

  5. SoldLabRecruit sold the pipeline I built to another agency.

    Data and AI pipelines

    Your most valuable data is scattered across systems no one can search.

    Pipelines that find, match, enrich and index it, with search on top.

    Four to ten weeks

I still ship code.

A strategist who still ships code. The work gets built, and you can see it.

SunMonTueWedThuFriSatNovDecJan ’26FebMarAprMayJunJulAugSepOct
9,804 contributions on GitHub in the last year. Each line is a day of the week; each rise is a week of work. Gold lines are weekends. github.com/quesurifn

How it works

  1. A conversation. You tell me what you’re building and where it hurts. If I’m not the right fit, I’ll say so and point you to someone who is.
  2. A one-page proposal. Scope, what you’ll have at the end, timeline and price.
  3. The work. Weekly written updates. You see what I see as I see it.
  4. A handoff. Documentation and walkthroughs, so your team owns what we built.

Or hire me full time.

Selectively open to senior roles and advisory work. Based in Phoenix, authorized to work in the US and Canada, remote or on site.

Read the record See the work

  • Head of product
  • Head of engineering
  • Head of AI
  • Staff or principal engineer

Tell me what you’re working on

I read every message myself and reply within two business days.

Have something urgent to say? call or text. Feel free to call. I prefer a text.