Training Turk

Mercor listing

Insurance Underwriting Specialists

$80/hr

The role in one line

An experienced underwriter shapes AI model training by creating realistic insurance scenarios and evaluating model outputs for technical accuracy and sound judgment.

Written by Training Turk from the public listing; it may be incomplete or out of date. Read the full posting on Mercor.

What you would do

  • Design authentic underwriting problems involving risk selection, coverage terms, exclusions, pricing logic, and endorsement scenarios
  • Write reference-quality underwriting decisions with clear rationale that models learn from
  • Score AI-generated responses using structured rubrics for accuracy in coverage, pricing, and judgment
  • Document errors found in model outputs, flagging misinterpretations, calculation mistakes, or reasoning gaps

Who they are looking for

  • Minimum 2+ years hands-on underwriting across any line of insurance (life, health, property, casualty, commercial, specialty, reinsurance)
  • Deep working knowledge of policy mechanics, rating, exclusions, and how coverage decisions connect to pricing
  • Ability to articulate complex underwriting reasoning in writing for AI researchers and non-underwriter audiences
  • Industry credentials preferred, such as CPCU, CLU, AINS, ARM, or AU; experience across multiple lines or specialty risks is a plus

Skills this role asks for

risk assessmentcoverage analysispolicy provisionsunderwriting judgmentpricing rationaleexclusions and endorsementsai evaluationdecision documentation

What the interview is likely to probe

  1. 1.Coverage and pricing logic

    The core of the role is creating scenarios and judging whether AI reasons correctly about coverage mechanics; superficial knowledge fails immediately.

    Expect something like: “A 35-year-old homeowner applies for coverage. Their property has a history of small claims. How would you structure a scenario showing the interplay between coverage limits, deductibles, and premium adjustment?”

  2. 2.Decision documentation

    Your written justifications train the model; vague or unclear reasoning produces poor training data and confuses non-underwriters on the research team.

    Expect something like: “You're writing a reference response declining coverage for a commercial client. What specific elements would you document to show both the underwriting logic and why that decision was appropriate?”

  3. 3.Evaluation rigor

    You must spot subtle errors in AI outputs; missing a flawed coverage interpretation or pricing mistake degrades the feedback quality the model receives.

    Expect something like: “An AI model says a policyholder's claim for business interruption loss should be paid under a property policy that explicitly excludes business interruption. How would you structure feedback to identify this error and explain why it matters?”

  4. 4.Scenario design diversity

    AI models learn from variety; creating only obvious or routine cases produces underdeveloped models that fail on edge cases and unusual combinations.

    Expect something like: “Design two underwriting scenarios you'd create for a medical professional specialty insurance program. Why would each one test different underwriting judgment?”

  5. 5.Cross-line expertise

    Insurance reasoning varies across lines; depth in one area plus breadth across others makes you more valuable for creating diverse, realistic training data.

    Expect something like: “You have underwriting experience in property. If asked to create scenarios in health insurance, what specific knowledge gaps would you flag, and how would you address them?”

Exercise you may get

Create 3 underwriting scenarios across different risk types or coverage questions. For each, write a reference decision with full rationale and identify 2 potential errors an AI model might make.

How to prepare

  • Review recent insurance market trends or regulatory changes in your specialty; be ready to explain how they reshape underwriting decisions
  • Prepare 2-3 examples of underwriting decisions you've made where the reasoning was non-obvious or involved balancing competing factors
  • Identify 2-3 common mistakes you've observed in underwriting or claim analysis that would signal a model's misunderstanding
  • Reflect on insurance scenarios where coverage terms, exclusions, or pricing create genuine complexity for decision-making

Facts

Pay
$80/hr
Commitment
hourly
Hours
40 per week
Work arrangement
remote · Remote
Posted
9/21/2026