What the work involves

This is deck work, not memo work. You will be handed prompts that mirror real engagement deliverables — a step-plan slide for a cross-border restructuring, a one-page summary of an uncertain tax position for an audit committee, a client-facing read on proposed legislation, a provision walk for a controller — and asked to produce the reference version, then evaluate what the model produced against it. Some tasks are generative (you build the gold-standard deck). Some are comparative (two model outputs, you rank them and say precisely why). Some are surgical (the model's structuring slide is 80% right; you mark the step that would not survive review and explain the consequence).

The evaluation bar is partner-review, not technically-defensible. A slide that is accurate but buries the answer on line four, or that presents an aggressive position without flagging it as such, or that mislabels a flow-through as a disregarded entity in the org chart, fails — and your written rationale has to name the failure in terms a non-tax annotator can act on.

What the platform screens for

  • Verifiable practice history. Which firm, which service line (M&A tax, transfer pricing, ASC 740, SALT, private client), what you actually owned on deliverables versus reviewed.
  • Depth under follow-up. The AI screener will push on a technical claim you make. Vague answers collapse; specific ones — the code section, the reason the structure was chosen over the alternative, what the client pushed back on — hold.
  • Slide craft as a distinct skill. Being a strong tax technician is necessary and not sufficient. Expect questions about how you structure a step-plan slide, how you handle a caveat that cannot fit on the page, what you do with a deck that has three messages fighting for the headline.
  • Calibrated critique. Can you separate wrong from stylistically different, and rank severity rather than listing every flaw at equal weight?

Logistics

Fully remote and fully asynchronous — tasks are claimed from a queue, with turnaround windows rather than fixed hours. Commitment is flexible between 5 and 20 hours per week, more if you want volume, and contributors commonly stack it alongside full-time practice. The observed rate on this listing is $80/hour; rates on this platform vary by task type and calibration performance and are not guaranteed. The interview process includes an AI voice screen.