Record a short script in your natural Bulgarian-influenced English accent, using a smartphone or computer with a microphone. Follow the supplied recording guide and submit your audio through the designated platform. The recording takes about 15 minutes or less. The project welcomes applicants of all genders and accepts only 3 to 5 speakers for this accent.
As an Employment Lawyer on a project to improve AI legal reasoning, you’ll assess employment law scenarios and review AI-generated analyses, arguments, and documents. Your legal judgment and written feedback will help identify errors and gaps and guide improvements to model outputs. Applicants must be based in the US, hold a U.S. law degree, and have an active bar license.
Mercor keeps this listing open to build a pool of mathematicians for future AI training and evaluation projects; it does not represent one active opening. You may write research- or competition-level problems with solutions, assess model-generated proofs, and identify gaps in reasoning. After you apply, a short AI interview takes about 20 minutes. Matching projects may start within a week or several months.
Create practical PyCharm tasks that reflect Python development work, from project setup and package installation to debugging, Git use, and test runs. Prepare the files and environment, record each workflow, and explain its context in SuperAnnotate. The work supports AI training by showing how developers handle real tasks in a Python IDE.
Create research-grade AI evaluation tasks from computational work you have performed in your field. AfterQuery commissions tasks across life, physical, Earth, mathematical, and engineering sciences. You will define how evaluators should assess a strong solution, revise your task after expert feedback, and may also review work from other domain experts. The work follows a flexible, task-based schedule with no fixed weekly hours.
Work on a United States-based customer project that uses real-world digital tasks to improve AI systems. You’ll carry out workflows on online platforms, use AI tools, assess their results, and record what happened through screenshots and structured feedback. Some assignments may involve purchases, payments, or cancellations. Checkout tasks require a device with fingerprint or face recognition; biometric data is not stored.
Use Android Studio on your own Mac running macOS and capture screenshots that exceed 2.5 megapixels. You need access to the software today; the company does not supply licenses, virtual machines, or hardware. Meet the image-size requirement with one qualifying high-resolution display or two monitors captured together as a single image.
Review AI-generated explanations and build reference answers across enzyme kinetics, metabolism, protein structure, molecular biology, and laboratory methods. You will check scientific claims, calculations, experimental context, and safety, then compare model responses and provide specific written feedback. The role is remote and contract-based, with work conducted in English.
Create and refine consulting challenges that help train AI models to handle real client work. You will build authentic business deliverables, write prompts for the analysis behind them, and assess model responses against consulting methods and standards. The role calls for at least six months of consulting experience, native English, and strong Excel and PowerPoint skills.
Edit real-world video content as part of a customer project that helps train AI systems. You’ll shape raw footage into finished pieces, apply creative and technical judgment, and share video expertise that can inform model development. Your background in video editing matters more than prior AI experience. The role is remote and contractor-based.
Listen to AI-generated audiobooks in US and North American English and assess whether the narration sounds natural, accurate, and enjoyable. You’ll mark speech and audio issues in review tools, identify where the listening experience falters, and rate listener satisfaction. The role calls for native or near-native English fluency, regular audiobook listening, and availability for about 20 hours each week.
Lead quality reviews for Assamese AI training projects by checking contributor work against project rubrics and language standards. You will guide trainers and QA reviewers, resolve questions about meaning and cultural context, and maintain resources such as style guides and onboarding materials. The role is remote and contract-based; selection includes an AI interview, a domain-specific task, and a recruiter interview.
Record a short script in your natural Slovenian-influenced English accent, following a step-by-step guide, then submit the audio through the designated platform. The recording takes about 15 minutes or less. The project accepts 3 to 5 speakers for this accent, and all genders may apply. Imitated accents do not qualify.
Work on projects that assess and improve advanced AI systems by applying Python and expertise in biology or a related field. You will create and solve research problems, analyze scientific data, and check computational results. The contractor role averages about 15 hours per week, with flexible days and hours, including weekends if you choose. Pay is output-based per task that meets project specifications, and task time varies. Selected experts should start their first task within 24–48 hours after onboarding.
Bring your mechanical design experience or postgraduate engineering studies to a research project with a leading AI lab. You’ll complete and explain real CAD workflows on your own Mac, producing training data that helps benchmark and improve AI models. The role calls for about 10–20 hours each week and requires a Mac running macOS 15 or later.
Create practical design workflows in Canva, then record your computer-based process and document the task in SuperAnnotate. Projects may cover social posts, presentations, flyers, infographics, brand assets, or export and sharing steps. You need B2-level English, working computer and browser skills, and the ability to prepare assets and references before recording.
Record a short provided script in your natural Punjabi-influenced English accent, using a smartphone or computer with a microphone. Follow the recording guide and send your audio through the designated platform. The recording should take about 15 minutes or less. The project welcomes all genders and will accept only 3 to 5 speakers for this accent; imitated accents will not be paid and will be removed from the project.
Evaluate AI-generated images by comparing them with original images you provide. You’ll look for visual differences and flaws in areas such as lighting, texture, proportions, and overall quality, then record your observations in clear written feedback. The project requires basic familiarity with AI-generated images, but not specialized AI experience. The role is remote and limited to people based in the United States.
Use your research expertise to assess mathematical reasoning in AI model outputs and shape the standards used to judge correctness. You will write rigorous problem specifications, reference solutions, and proofs, then build evaluation sets in your area of specialization. The role is fully remote and requires a United States base.
Lead Japanese quality reviews across AI training projects, checking model outputs and contributor work against project rubrics. You’ll give specific feedback, answer language questions, coordinate updates through Discord, and maintain resources such as Japanese style guides and QA trackers. This is a remote contract role with no immediate project; qualified candidates may be contacted when relevant work becomes available.
Location and eligibility
Remote · Eligible in 7 countries
Engagement type
Contract
Compensation
Up to $55/hr
japanese
english
japanese quality assurance
Language & LinguisticsView job
Showing 20 active roles of 1,907. Dates reflect what the source published, not when AIWorkJobs last synced.
Sources: ten specialist AI work platforms. AIWorkJobs does not verify every claim made by a source. Report a listing that looks wrong through the contact page.