This description is a summary of our understanding of the job description. Click on 'Apply' button to find out more.
Role Description
This role involves enhancing large-language-model performance on realistic reasoning tasks. Experts will:
-
Write challenging, creative, and realistic tasks for AI models to solve
-
Write gold-standard solutions to those tasks
-
Evaluate AI-generated responses for clarity, accuracy, and comprehensiveness
Qualifications
-
Experience in one or more of the following archetypes: technical writers, editors, research analysts, academic editors, content reviewers, legislative analysts, reading-comprehension specialists, assessment authors, curriculum designers, or instructional designers
-
Demonstrable experience creating Q&A, summaries, or problem-solving tasks for long-form content
-
Ability to excel at deep reading and quickly synthesize information across a variety of domains
-
Strong fact-checking and editing skills to validate and polish AI-generated answers
-
Analytical judgment to spot inconsistencies, gaps, or misinterpretations in draft outputs
-
Clear and concise written communication with high attention to detail
Requirements
-
Collaborate directly with Mercor coordinators and the client’s research leads
-
Availability of up to 40 hours of work per week
-
Fully remote and asynchronous work, flexible to your schedule
-
Project length is scoped to last a minimum of 6 weeks with potential extensions
Pay and Legal Status
-
Legally classified as an hourly contractor for Mercor
-
Payment at the end of each week via Stripe Connect
Screening Process
-
Complete a short interview and form, taking a total of 20-30 minutes
-
Participate in a paid work trial (4 hours) to complete domain-specific tasks
-
Currently accepting applicants from the U.S., UK, Canada, and Australia
Company Description
Mercor specializes in recruiting experts for top AI labs and is based in San Francisco, CA. Our investors include Benchmark, General Catalyst, Peter Thiel, Adam D’Angelo, Larry Summers, and Jack Dorsey.
