Mercor

Japanese

• Open and check provided PDF tasks
• Confirm pages are in Japanese, legible, contain real content, and show no personal details
• Identify and bound every meaningful page region
• Assign component types and reading-order indices
• Link regions to associated figures or tables using parent component identifiers
• Transcribe all text faithfully, including kanji, hiragana, katakana, furigana, and handwritten content
• Flag illegible source regions
• Capture language, document type, source, page dimensions, and table, formula, and handwriting flags
• Review colleagues' annotation and transcription work as an experienced annotator
• Produce human-authored structural annotations and transcriptions for AI training data

PDF Annotation, Transcription Expert – Japanese

• Open and check provided PDF tasks
• Confirm pages are in Japanese, legible, contain real content, and show no personal details
• Identify and bound every meaningful page region, including titles, headings, paragraphs, lists, tables, figures, diagrams, captions, formulas, questions, and answer fields
• Assign component types and reading-order indices
• Link regions to related figures or tables using parent component identifiers
• Transcribe all text exactly as shown, including kanji, hiragana, katakana, furigana, and handwritten content
• Flag illegible source regions
• Capture page metadata, including language, document type, source, dimensions, and table, formula, and handwriting flags
• Review colleagues' work end to end as an experienced Japanese expert
• Ensure regions are correctly captured, bounded, typed, and ordered
• Ensure transcriptions match the source character for character and pass second-expert review

Safety Specialist

• Write challenging single-turn prompts in the hazardous-device/public-safety domain
• Label prompts as benign, dual-use, or adversarial
• Evaluate model responses against a defined policy standard
• Judge whether responses were handled correctly
• Write reference answers describing correct responses and the technical reasoning behind them
• Red-team frontier AI models for misuse-potential assessment
• Sustain reading and writing about misuse scenarios in the field

Hazardous Device Specialist / Public Safety Professional

Artificial Intelligence

• Red-team conversational AI models and agents through jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation
• Generate human data by annotating failures, classifying vulnerabilities, and flagging systemic risks
• Apply taxonomies, benchmarks, and playbooks to keep testing consistent
• Produce reproducible reports, datasets, and attack cases for customers
• Review AI outputs involving sensitive topics such as bias, misinformation, or harmful behaviors
• Help expand evaluation coverage and strengthen customer AI systems
• Collaborate with leading researchers on projects focused on training and enhancing AI systems

AI Safety Expert – English, Assamese

Engineer

• Write challenging single-turn prompts in the chemistry and chemical safety domain
• Label prompts as benign, dual-use, or adversarial
• Evaluate model responses against a defined policy standard
• Judge whether responses handled technical requests correctly
• Write reference answers and explain the technical reasoning behind them
• Distinguish legitimate professional questions from potentially dangerous misuse requests
• Provide written rationales understandable to non-specialists
• Read and write about misuse scenarios in the field for sustained periods

Industrial Hygienist / Process Safety Engineer

iOS Engineer

• Use Excel on macOS for spreadsheet modelling, formulas, pivot tables, and charting
• Work on projects training and enhancing AI systems
• Provide human expertise to Mercor's AI lab and enterprise partners
• Collaborate with leading researchers
• Complete projects on an independent schedule; projects may be extended, shortened, or concluded early based on needs and performance

Application Users – Excel, macOS

• Red-team conversational AI models and agents through jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation
• Generate human data by annotating failures, classifying vulnerabilities, and flagging systemic risks
• Follow taxonomies, benchmarks, and playbooks to keep testing consistent
• Produce reproducible reports, datasets, and attack cases
• Review AI outputs involving sensitive topics such as bias, misinformation, and harmful behaviors
• Identify vulnerabilities automated tests miss
• Expand evaluation coverage and reduce production surprises
• Support Mercor customers in making AI systems safer and more trustworthy

AI Safety Expert, English, Odia

• Work on projects focused on training and enhancing AI systems
• Apply Excel expertise, including spreadsheet modelling, formulas, pivot tables, and charting
• Contribute human expertise to projects for leading AI labs and enterprises
• Collaborate with leading researchers
• Complete work remotely on a self-directed schedule
• Perform services according to project needs and performance expectations

Application User – Excel, macOS

Financial Planning and Analysis

• Review and QA AI-agent rollouts in Financial Forecaster
• Evaluate completed agent runs against the actual model and documents
• Verify the correct scenario, account, coordinate, basis, conclusion, and deliverable
• Identify agent failure modes and explain why agents succeed or fail
• Flag tasks that are too easy or solvable without using Financial Forecaster
• Refine task prompts and seeded worlds to reflect authentic FP&A and lender-reporting work
• Increase task realism and difficulty for capable AI agents
• Update grading guidance so correct answers pass and incorrect answers fail
• Detect gating failures such as wrong scenario or vintage, wrong accounting basis, excluded add-backs, unwritten model corrections, or conclusions given away by source documents
• Catch plausible but incorrect answers stated at the wrong grain or inconsistent with the reporting pack
• Work on projects training and enhancing AI systems for Mercor and its research partners

FP&A Expert

• Design adversarial prompts to stress-test frontier AI models
• Identify jailbreaks, unsafe behaviours, hallucinations, and policy failures
• Evaluate model robustness across misinformation, cyber, biosecurity, fraud, political content, and other sensitive domains
• Document vulnerabilities and contribute to safety benchmarking and red-teaming reports
• Collaborate with AI researchers to improve model alignment, robustness, and safety
• Work on projects focused on training and enhancing AI systems

AI Safety Red Teamer

• Red-team conversational AI models and agents using jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation
• Generate human data by annotating failures, classifying vulnerabilities, and flagging systemic risks
• Apply taxonomies, benchmarks, and playbooks to maintain consistent testing
• Produce reproducible reports, datasets, and attack cases for customers
• Review AI outputs involving sensitive topics such as bias, misinformation, and harmful behaviors
• Expand evaluation coverage and uncover vulnerabilities missed by automated tests

AI Safety Expert, English – Bengali

General

• Write challenging single-turn prompts in the health physics domain across benign, dual-use, and adversarial levels
• Evaluate model responses against a defined policy standard
• Judge whether responses handled technical requests correctly
• Write reference answers explaining correct responses and the technical reasoning behind them
• Distinguish routine professional questions from requests indicating potential misuse
• Apply practical experience in survey, assay, contamination control, and work planning to red-team frontier AI models
• Provide written rationales that non-specialists can follow
• Read and write about misuse scenarios in health physics for sustained periods

Health Physicist

• Red-team conversational AI models and agents through jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn manipulation
• Annotate failures, classify vulnerabilities, and flag systemic risks
• Apply taxonomies, benchmarks, and playbooks to maintain consistent testing
• Produce reproducible reports, datasets, and attack cases
• Probe AI outputs involving bias, misinformation, and harmful behaviors
• Help train and enhance frontier AI systems for Mercor’s AI lab and enterprise customers

AI Safety Experts – English, Odia

• Write challenging single-turn prompts in industrial hygiene, chemistry, and chemical safety, labelled as benign, dual-use, or adversarial
• Evaluate model responses against a defined policy standard
• Judge whether each response was handled correctly
• Write reference answers explaining what a correct response looks like and the technical reasoning behind it
• Red-team frontier AI models for misuse potential in technical requests
• Produce written rationales that non-specialists can follow
• Read and write about misuse scenarios in the field for sustained periods