What the work involves

You will spend most of your day inside two artifacts: a prompt describing a business scenario, and a slide deck — sometimes model-generated, sometimes human-drafted — that attempts to answer it. Your job is to judge whether the deck actually does the job a partner would expect: is the storyline sound, does the action title state a conclusion rather than a topic, is the supporting evidence on the slide consistent with the title, does the chart type fit the data, and is the formatting clean enough to put in front of an executive committee. Where it falls short, you rewrite rather than simply flag — refined slide content, corrected structure, and written rationale explaining what was wrong and what "good" looks like.

Expect a mix of task types across a typical engagement: rewriting weak action titles, restructuring a deck that buries the recommendation, correcting numeric or logical inconsistencies between narrative and exhibit, sharpening prompts so they elicit better outputs, and comparing two candidate decks and arguing which is stronger. The rationale you write is often more valuable to the client than the corrected slide, because it is what teaches the model.

What the platform screens for

  • Verifiable deck history — where you built decks, for whom, at what level of audience, and over how many years. Vague "I've made lots of presentations" answers do not survive follow-up.
  • Structured thinking you can name and apply — pyramid principle, MECE, so-what titling, horizontal and vertical logic. Screeners probe whether you use these in practice or only recognise the vocabulary.
  • Judgment under ambiguity — what you do when the prompt is underspecified, when the data contradicts the stated conclusion, or when two decks are flawed in different ways.
  • Written precision — your written answers are themselves a work sample; rambling prose is read as a predictor of rambling slide copy.

Logistics

Fully remote, contractor engagement of up to ten weeks. Turing states a commitment of at least 8 hours per day, up to 40 hours per week, with at least 4 hours overlapping Pacific time — this is not a nights-and-weekends side project, and candidates who cannot hold the overlap window are routinely filtered out. A bachelor's degree or equivalent practical experience is expected; prior AI evaluation or annotation work is welcomed but not required. Pay is undisclosed on this listing; ask for the rate and the task-volume expectation before signing, and confirm whether hours are guaranteed or drawn down against available task supply.