What the work actually involves
You will spend most of your time in three modes. First, authoring reference deliverables: a protocol synopsis from a target product profile, an efficacy summary table with the analysis populations defined correctly, a medical affairs deck that survives MLR review. Second, evaluating model output against those standards — reading an AI-drafted CSR Section 12 and marking where it conflates ITT and per-protocol results, invents a footnote, or states an efficacy claim the data does not support. Third, writing the rationale, which is where the real value sits: a note explaining why an investigator brochure summary is unacceptable under ICH E6 is worth more to the lab than a pass/fail flag.
Tasks arrive as structured briefs with file artifacts — .docx, .xlsx, .pptx. Expect real spreadsheet work: cross-tabs, formula integrity, table shells that match a statistical analysis plan. Expect real slide work: a claim, a source, a figure that supports it. Sloppy craft in the source files is itself a failure mode the lab is trying to eliminate.
What the platform screens for
Ethos runs an AI voice screen before human review. It probes for direct authorship — not adjacency. Saying you "worked on regulatory submissions" invites an immediate follow-up about which CTD module, whether you drafted or reviewed, and what the reviewer feedback was. The screen also tests evaluation judgment: given a plausible-sounding but subtly wrong AI output, can you name the specific defect and rank it against other defects? Candidates who describe process without demonstrating standards tend not to advance.
Logistics
- Fully remote, fully asynchronous — no standing meetings, no fixed timezone
- 5–20 hours per week, self-scheduled; higher volume available to reliable contributors
- $100/hour is the rate observed on this listing, not a guarantee; rates vary by task complexity and reviewer tier
- Contractor engagement; work is delivered against briefs with turnaround windows rather than live coverage