AI Researcher
First 90 days
Research how AI answers recommend brands, use sources, and change over time. Turn that work into better measurement.
- Build an initial evaluation map for the core prompt-answer-citation workflow.
- Identify the highest-value reliability gaps in current analysis outputs.
- Ship at least one improvement that materially raises trust in reported AI visibility signals.
Why this role exists
This role sits at the boundary between research, product judgment, and system design. You will help define what high-quality AI visibility analysis actually means, design evaluation methods that survive contact with reality, and turn that work into product behavior teams can trust.
The category is moving faster than the measurement standards around it. We need someone who can turn vague claims about AI visibility into rigorous, usable, product-grade judgment.
What you will work on
- Design evaluation frameworks for answer quality, citation behavior, positioning, and sentiment.
- Work closely with product and engineering to turn research insights into concrete product logic and operator-facing workflows.
- Investigate failure modes in extraction, normalization, and interpretation across assistants and prompt types.
- Build the internal research habits, benchmarks, and review standards that keep the product credible as models change.
What a strong fit looks like
- A strong research instinct paired with the ability to operationalize findings into product or system behavior.
- Comfort reading raw model output closely and separating signal from noise without hand-waving.
- Evidence of work in applied LLMs, information retrieval, evaluation, search relevance, or adjacent research problems.
- Clear writing and the ability to explain why a method is trustworthy, not just interesting.
What will excite you here
- Turning ambiguous AI behavior into repeatable evaluation logic.
- Working across product, engineering, and research instead of staying confined to a narrow experimentation lane.
- Building the measurement layer for a category that still lacks strong standards.
First 90 days
- Build an initial evaluation map for the core prompt-answer-citation workflow.
- Identify the highest-value reliability gaps in current analysis outputs.
- Ship at least one improvement that materially raises trust in reported AI visibility signals.
Interested in the work?
Chatobserver is not accepting applications for these roles right now.
careers@chatobserver.comAbout the company