What the work actually involves
You author self-contained design problems and then judge whether a model's answer holds up. A task might ask a model to fix a slide where the message is buried under six competing visual elements, choose a chart type for a dataset where the obvious choice misleads, or specify build timing for a phased reveal that supports a spoken narrative rather than distracting from it. Each item needs a defensible reference answer and a rationale that a second reviewer can follow — 'this looks better' is not a rubric. When you grade model output, you write feedback that names the specific principle violated: hierarchy inverted, colour used to decorate rather than encode, axis truncated, type scale collapsed to a single weight.
What the platform screens for
AfterQuery is looking for people who have shipped real decks under real constraints — brand systems, executive audiences, sales cycles — not just people who know PowerPoint's menus. Screens tend to probe whether you can articulate why a design choice works, defend it against a plausible counter-argument, and separate defensible principle from personal taste. Depth in advanced PowerPoint (master slides, layouts, morph, animation sequencing, chart formatting, template architecture) is checked directly. Data-visualisation judgement and the ability to write clearly about visual decisions carry significant weight, since most of the deliverable is prose about images.
Logistics
- Fully remote and asynchronous — no fixed hours, no standing meetings in most engagements.
- Volume is project-driven; contributors commonly report 5–20 hours a week, with more available during active batches.
- Expect written feedback loops with reviewers and iteration on task difficulty and clarity.
- Adjacent tool experience (Figma, Adobe CS, Keynote, Google Slides) is a plus, not a gate.
- Pay observed at roughly $50/hr; rates vary by project and are not guaranteed.