Turing is hiring healthcare administrators and clinical informatics specialists to build EHR task environments in OpenEMR — patient registration, encounter charting, e-prescribing, and claim workflows — that AI agents are then measured against. You execute each workflow yourself, record the reference trace, and define the database and state checks that decide whether a model's attempt was actually correct.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually involves

This is environment authoring for agentic AI evaluation, using a live OpenEMR instance as the workbench. A typical task starts with a clinical administration scenario you invent and make plausible: a new patient registration with insurance on file, an encounter backlog to be charted, a diagnostic order to place and resolve, a prescription to write, or a claim to build and push toward a clearinghouse. You then produce the supporting mock assets — patient profiles, lab-order briefs, payer details, claim files — and execute the full workflow in OpenEMR yourself, capturing a reference execution trace.

The harder half is verification. After executing, you specify exactly what a correct end state looks like: which tables and fields changed, which encounter and billing statuses are valid, what the claim ledger should read, what the medical record must contain. Checks need to be tight enough that a plausible-looking but wrong model run fails, and loose enough that a legitimately different correct path still passes. You will also review other contributors' clinical task instructions for ambiguity, coding errors, and workflows that cannot actually be completed as written.

What the screen looks for

Real operational exposure to an EHR — OpenEMR hands-on is stated as mandatory, with substantial MEDITECH, athenahealth, Epic, or Cerner experience as the accepted alternative.
Fluency in the patient encounter lifecycle end to end, not just one desk within it: registration, documentation, orders, prescribing, coding, claim submission, denial and rework.
ICD-10 and CPT working knowledge, and HIPAA conventions applied to synthetic data (no real PHI, ever).
Whether you can say why a record or billing state is technically correct, citing the field or table rather than describing how the screen looked.
SQL against MySQL/MariaDB, and HL7/FHIR familiarity, are preferred rather than required — but they raise your ceiling on verification work.

Logistics

Fully remote, hourly, with a stated five-week duration and possible extension based on project needs. Pay is undisclosed on this listing; Turing rates are typically negotiated per engagement and per seniority, so treat any figure you hear secondhand as unverified. Work is largely asynchronous against task quotas and review deadlines, with occasional calibration calls; expect a start as soon as onboarding clears.

Domain Expert - Health/EHR Administrator

What the work actually involves

What the screen looks for

Logistics

Interview prep — 10 questions