What the work involves
You'll work on presentation and deck tasks the way an associate actually experiences them: take a prompt or a fact pattern, produce the storyline, then write or rebuild the pages. A typical task might be drafting an executive summary page for a sell-side process, restructuring a model output into a clean exhibit, or reviewing a model-generated deck and marking exactly where the horizontal and vertical logic breaks down. Much of the value sits in the critique — explaining why a page fails, not just that it does, in language precise enough that a model can learn the distinction between a defensible action title and a descriptive label.
Expect a mix of generation and evaluation. Some weeks lean toward producing reference decks; others toward side-by-side comparisons of model outputs, rubric scoring against dimensions like MECE structure, exhibit choice, and client-readiness, and written rationales that hold up under review. Ethos and the lab both audit for consistency, so your scoring needs to be reproducible rather than instinctive.
What the screen looks for
The AI voice screen verifies the résumé facts first — firms, group, years, seniority at each stage — then goes deep on deck craft. Expect follow-ups on how you'd fix a specific weak page, what makes an action title wrong, when a waterfall beats a bar chart, and how you'd handle a model output that is polished but subtly misleading. Vague references to "banker polish" collapse quickly under a second question. Candidates who can name their reasoning — the pyramid, the so-what test, the discipline of one idea per page — do well.
Logistics
- Fully remote, fully asynchronous; no standing calls
- Observed band around $100/hour, dependent on task type and calibration performance — not guaranteed
- 5–20 hours per week, scalable up if you have capacity
- Work is tracked per task; you choose when to pick it up