AI Safety Data Annotator / Agentic Trajectory Author

The title combines two things: annotating data for AI safety, and authoring agentic trajectories. The post describes the writing side in most detail. It wants someone who can produce believable back-and-forth conversations in which one side behaves like an actual customer.

Beyond conversation, the post mentions tool calls, function outputs and structured data. So the writing is not only prose; it also covers the technical pieces that sit alongside a conversation. Python is named as a working skill, and documentation is listed as well.

The post does not say how the work is divided, which team you would sit in, or what tools are used day to day.

• Education: a bachelor's degree, in any discipline.
• Writing: strong English, enough to write realistic dialogue and structured content.
• Thinking: logical reasoning.
• Python: good working knowledge.
• Soft skills: good communication and documentation.

Freshers are explicitly welcome if the writing and reasoning are strong. Hopin's data puts the experience range at 0 to 4 years. The role is remote.

• It helps to write a few sample customer conversations of several turns each, so you have something concrete to show your writing range.
• Practise representing information in structured formats using Python, such as dictionaries or JSON-style data, since structured data is part of the ask.
• Try writing a short example of a function being called and the output it returns, then check that the conversation around it stays consistent.
• Keep brief notes on why you made each writing choice. Documentation is a listed skill, and clear notes show it.
• Reread your samples for logical gaps, because reasoning is named alongside writing.