What the work actually involves
AfterQuery contracts historians to build evaluation items in their own period and region of expertise. A typical unit of work starts from a primary source or a contested causal question — a parish register, a treaty draft, a set of trial depositions, a disputed dating — and turns it into a scenario where a strong answer requires reading the evidence rather than recalling a summary. You then write the reference answer, and the reference answer is the substance of the job: not just the correct interpretation but the chain of inference, the source limitations, the alternative reading a competent scholar would consider, and why it loses.
The grading side is the other half. You review model output for factual accuracy, anachronism, misattributed causation, and the specific failure mode the brief names directly — prose that reads authoritative while flattening the evidentiary record or inventing causal links no historian would assert. Flagging that convincingly means quoting the model, naming what it did, and pointing at what the sources actually support.
What the screen looks for
- A real specialty with primary-source hands on it. Expect follow-ups on archives you have used, languages you read, and what a specific source type can and cannot support.
- Historiographical awareness. Whether you can articulate a live debate in your field and represent both positions fairly inside a single task.
- Calibration. Willingness to mark a plausible, well-written answer wrong, and to distinguish an interpretive difference from an error.
- Written clarity. Reference answers are read by non-historians; reasoning has to survive that.
Logistics
Fully remote and asynchronous, no set meetings, minimum roughly 10 hours per week with volume you scale up or down. Pay is weekly via Stripe; the $50–100/hr band is as observed on the platform and the individual rate is set from credentials and experience rather than negotiated per task.