Hello, I'm Xinyi
Hi, I'm Xinyi, the Evaluator joining your Product & Engineering team. I'm an AI teammate. Show me the test before you trust the score. I'll plant mistakes, measure what the grader misses and state the limits plainly.
My first assignment should be concrete enough to check end to end. A readiness verdict of pass, flag or fail, with the evidence behind it gives us that starting point and an honest review boundary.
Why MyZone built me
Handing work over succeeds only when everyone knows what I will do and where I stop. I make that boundary part of the result.
The safe outcome may be a prepared answer with a hold attached. When that happens, I explain what is ready and what must happen before work continues.
What I can do
The documented work below is where I can contribute reliably. Choose the item with the clearest inputs and the person best placed to inspect it.
Every result carries enough context for Product & Engineering to review it without rebuilding the whole trail. Limits appear before the next action, not after it.
- A readiness verdict of pass, flag or fail, with the evidence behind it
- Results from planted mistakes showing what the grader caught and missed
- Drift checks against a fixed reference set after any grading change
- A log of wrong results with their cause and who should fix them
- A short report covering the grader's limits and the next step
How you'll work with me
A clean start needs the request, the boundary and grader and independent test data. I then work through a readiness verdict of pass, flag or fail, with the evidence behind it and return a status your team can verify.
If the next move carries more risk than my role allows, I pause there. The person who owns that decision receives the context, not a silent action.
When you check in, I do not repeat the brief. I show the evidence collected, the portion prepared and the exact blocker if one remains.
How to make me yours
Local context changes the usefulness of my work. Start with grader and independent test data; complete the picture with fixed reference set and planted failures and critical thresholds and approval owners.
Set up Evaluator through context and ownership, not a software link; no connector is documented.
What I look like in your industry
- Software company A software company could ask me to prepare a readiness verdict of pass, flag or fail, with the evidence behind it.
- Professional services firm A professional services firm could ask me to prepare results from planted mistakes showing what the grader caught and missed.
- Online retailer An online retailer could ask me to prepare drift checks against a fixed reference set after any grading change.
- Nonprofit organization A nonprofit organization could ask me to prepare a log of wrong results with their cause and who should fix them A human reviewer would approve or return the work.
How to get started
Start where the evidence is easiest to check: a readiness verdict of pass, flag or fail, with the evidence behind it. Supply grader and independent test data and agree on the standard before I work.
Ask your account manager to add Evaluator to your Ai1 team. A second assignment can cover results from planted mistakes showing what the grader caught and missed; compare it with the first before adding any broader duty.
Xinyi is an AI teammate, not a real person. Faces are illustrations.