Government experts, making AI work for government.

Your Roster is building the largest network of government experts to give AI models the human feedback they need. People who have run the programs, written the contracts and briefed the Hill judge what models get right, what they get wrong, and why.

Model answer under review
“A sole-source justification can cite urgency alone if…”
Experts judgingReasoning recorded

A model can pass a benchmark and still get a contracting question wrong. The people who can tell are the ones who have done the work.

What experts do.

1Write the questionsReal tasks from your field, the kind a model will be asked on the job: a contract clause, a program decision, a policy memo.
2Judge the answersGrade what models produce against a rubric written for your domain. Where you are unsure, say so: unknown is a valid answer.
3Explain whyThe reasoning behind a judgment is what teaches a model. You talk it through; Your Roster records how you got there.

Who we’re looking for.

Mid-career public-sector and regulated-domain professionals with a real track record, in government now or recently.

Current federal employees can take part within their agency’s rules. Unpaid work needs nothing on file. Paid work needs your agency ethics office’s written approval, recorded with us first.

Program managers
Contracting and procurement officers
Policy analysts
IT specialists
Congressional staffers
Compliance and regulatory affairs

How joining works.

Join
Tell us your field, your agencies and the work you’ve done.
Get verified
Employment and credentials checked, and colleagues who worked beside you vouch for it.
Get matched
We invite you to engagements that fit your domain, after a conflict check.
Do the work
Write, judge and explain, remotely, on your schedule. Paid where your rules allow.

For AI labs, integrators and program offices.

Book a verified expert cell: five to twenty professionals matched to your domain, screened for conflicts, working to a rubric you agree on. Sold as a cell, never as headcount.

Any network can say an expert passed its assessment. Your Roster can tell you who worked alongside them.

Request an introduction
Every result comes with its evidence
RequestWhat the model was askedResponseWhat it saidJudgmentsEach expert’s call and reasoningNot evaluatedShown as such, never as a passHashSHA-256 of the whole record

Run on Aegis Eval, our evaluation engine. No single trust score stands in for the evidence.

Who’s building it.

Savan Kong
CEO & Co-Founder
DoD’s first Customer Experience Officer; Defense Digital Service; Redfin, Amazon Kindle.
Matt Kelly
Co-Founder, Growth
Dave Spellman
Co-Founder, COO
20+ years scaling service delivery at Extensiv and Accruent.
Christan Johnson
Co-Founder, Experience & Community
20+ years of executive recruiting.
Morgan AudinoAdvisor. Head of Strategic Projects, Handshake AI
Jason PickartAdvisor. Office of the DoD CIO; President, NCMA Dayton
Philip ReimanAdvisor. Former Lead Attorney, DoD CDAO; former General Counsel, Defense Digital Service

Questions.

Do I need AI experience?No. You need experience in your field. We show you how the work is done.
I’m a current federal employee. Can I join?Yes. Unpaid work needs nothing on file. For paid work, record your agency ethics office’s written approval with us first.
Will I work with sensitive information?No. Nothing classified, no CUI. The work draws on your judgment, not your access.

Your judgment, on the record.