Expert sourcing layer for AI evaluation

The hard part of an eval isn't the workflow.It's finding the experts.

For AI data and evaluation companies: verified, available experts for the credential-heavy verticals where internal sourcing stalls. At project speed.

Talk through your hardest vertical See how the search works
§ 01 · Method

Search by evidence, not pedigree.

Most sourcing starts with credentials and keywords. We start with the work - published papers, repos, case records, published results - and verify expertise before a name ever reaches your team.

01
Map

Break the vertical into signals

We turn a hard vertical into the signals that prove real depth - published work, code, case records, the niche communities specialists actually live in. The things résumés and keyword search miss.

02
Search

Source from proof of work

We search the open web where experts leave evidence - papers, repositories, case records, published results, leaderboard standings. The scarce specialists rarely sit in a recruiter database.

03
Verify

Verify before you see a name

Credential fraud is real in this market. We confirm every claim against the evidence ourselves - depth, identity, availability - so the names you receive clear your screen, not just ours.

§ 02 · Example search

What you actually get.

Every candidate arrives with verified credentials, evidence of depth, and confirmed availability. You evaluate fit, not chase ghosts.

Candidate snapshot
Dr. M. Chen
Mathematical Reasoning Expert
PhD applied math · 8 years research · published on logic & reasoning
Interested
ReadinessReady this week
CompIn range
Why they fit

PhD with published work on chain-of-thought reasoning. Has already contributed to two prior RLHF projects.

Evidence

Published papers, eval frameworks on GitHub, active in math-reasoning communities.

Next step

Ready for a first conversation this week.

Our note

Strong fit for graduate-level reasoning data and formal-logic eval. Lighter on production ML, but this project won't require it.

§ 04 · Founders

Two founders. One inbox.

You always talk to the people running the search.

Daniel Kovari
Daniel Kovari
Co-founder · Formerly Head of Growth at a YC company
LinkedIn →
Bernat Nacsa
Bernat Nacsa
Co-founder · Formerly VC and VC-backed founder
LinkedIn →
§ 05 · FAQ

Frequently asked.

The credential-heavy ones internal teams stall on - competition-math PhDs, securities lawyers, attending physicians, quant traders, native low-resource-language speakers, ex-MBB consultants. We staff individual specialists and build RLHF and SFT contributor networks in these domains, from a first hire to a full bench.
Flat and transparent. A fixed fee per search, agreed up front. No percentage-of-salary, no surprise costs.
Credential fraud is real in this market, so we verify before you see a name. Every claim is checked against the evidence - published work, code, case records - alongside identity, domain depth, and current availability.
Typically 1-2 weeks from the intake call. A first shortlist of verified, available specialists, even in verticals most teams take a quarter to fill.
*
§ 06 · Start a search

Tell us the vertical you're struggling to staff.

We'll show you who's out there, and how fast we'd staff it.

Talk to us Free sourcing assessment · No commitment